Conventional Modalities and Novel, Emerging Imaging Techniques for Musculoskeletal Tumors Meera Raghavan, MD

Diffusion-weighted, spectroscopy,

and in- and out-of-phase imaging are

expanding the role of MRI for

characterizing musculoskeletal tumors.

Irena Orlov. Industrial feel 5687. Digital on canvas, 40" × 60".

Conventional Modalities and Novel, Emerging Imaging Techniques for Musculoskeletal Tumors Meera Raghavan, MD

Background: Imaging of musculoskeletal tumors requires a multimodality approach and includes radiography, computed tomography (CT), and magnetic resonance imaging (MRI). Methods: Topics related to primary bone and soft-tissue tumors are reviewed. The fundamental imaging principles are discussed as well as the applications of emerging imaging modalities. Results: MRI is the preferred technique for the evaluation of musculoskeletal tumors, whereas other imaging modalities play a complementary role. Radiography is indicated as the first-line imaging modality in bone and soft-tissue tumors, whereas CT is the preferred modality for evaluating cortical osseous lesions or calcifications and in patients with contraindications to MRI. Positron emission tomography (PET)/CT and PET/MRI are helpful in identifying the glucose metabolism of the lesion. Ultrasonography is the most useful for biopsy guidance and can aid in differentiating cystic from solid masses and identifying vascularity. Novel modalities, such as diffusion-weighted imaging, spectroscopy, and habitat imaging, show promise in increasing diagnostic accuracy and affecting treatment strategies. Conclusions: Conventional modalities and emerging, novel imaging techniques can provide noninvasive methods to diagnose and evaluate musculoskeletal tumors.

Introduction plain radiography, followed by a cross-sectional The imaging evaluation of musculoskeletal tumors modality such as computed tomography (CT) or involves a multimodality approach. Each modality magnetic resonance imaging (MRI).2 Nuclear medi- provides different diagnostic information. The pa- cine scintigraphy, ultrasonography, and positron tient’s history, physical examination findings, and emission tomography (PET) may also be performed location of the abnormality will determine the type in conjunction with or after the initial assessment. of imaging modality.1 The initial evaluation of bone MRI is the technique of choice for the imaging as- and soft-tissue tumors is typically performed with sessment of soft-tissue tumors, whereas CT or MRI can be used to determine the extent of disease in bone tumors. From the Department of Radiology, Northwell Health, New Hyde Park, New York. Novel techniques such as diffusion-weighted Address correspondence to Meera Raghavan, MD, Department of MRI, spectroscopy, and chemical shift are expand- Radiology, Northwell Health, 270-05 76th Avenue, New Hyde Park, ing the role of MRI in the characterization of muscu- NY 10040. E-mail: [email protected] loskeletal tumors. MRI-defined “habitats” in sarcoma Submitted March 23, 2016; accepted January 6, 2017. provide another novel and evolving method to char- No significant relationship exists between the author and the com- panies/organizations whose products or services may be refer- acterize and quantify features on MRI so as to provide enced in this article. prognostic and therapeutic information.

April 2017, Vol. 24, No. 2 Cancer Control 161 Conventional Modalities of soft-tissue masses has been supplanted by MRI.9 Plain Radiography Radiography does not obviate the need for cross-sec- Plain radiography remains the first-line choice of imag- tional evaluation and should be considered comple- ing modality for the initial evaluation of a bone lesion mentary to MRI and CT.1 following clinical examination.2 Radiographs can be Findings on radiography can also be used to tai- obtained quickly, easily, and at low cost. The differen- lor cross-sectional imaging. For example, if metal den- tial diagnosis of many primary bone tumors is based on sity is noted, then sequences that would normally be radiographic features.2,3 By combining radiographic fea- performed with fat saturation can be altered to short tures, such as lesion location, size, matrix, margin, and tau-inversion recovery imaging to mitigate artifacts.10 periosteal reaction, with patient age, sex, and clinical If tumor involvement of the adjacent bone is seen, then symptoms, the health care professional can effectively further cross-sectional evaluation can be performed narrow the differential diagnosis to a small number of with CT because it is more useful than other modalities possibilities (Fig 1). The bone lesion can be classified as in the evaluation of cortical bone. Results on radiog- nonaggressive or aggressive from its radiographic ap- raphy may also suggest that the process is articular or pearance. A classic, nonaggressive bone lesion may not juxta-articular (eg, synovial chondromatosis, tumoral require subsequent imaging or treatment unless further calcinosis); in that case, the imaging planes or sequenc- anatomical information is required and surgical interven- es that optimize the specific joint can be performed.11 tion is considered.4 Conversely, radiography can provide information about a lesion that helps determine appro- Ultrasonography priate further imaging with CT or MRI and useful clini- Ultrasonography can be complementary to MRI or CT cal information about the potential fracture risk of the le- in the initial evaluation of soft-tissue masses.12 In gen- sion.5 If multiple lesions are present on radiography, then eral, it is widely available, portable, is associated with the differential diagnosis will change to include other di- low cost, and can be performed without ionizing radia- agnoses such as metastatic disease or myeloma.6 tion.13 In particular, ultrasonography can be used for Radiographs are indicated in the workup of a soft- the initial evaluation of soft-tissue masses in patients tissue mass.1 Radiographs should be inspected for with contraindications to MRI. Sonography can con- tissue plane distortion, lucent areas — which might firm the presence of the mass, characterizing it as solid indicate fat — bony remodeling, and soft-tissue calcifi- or cystic, and assess its dynamic features such as com- cations or mineralization.7,8 For example, a peripherally pressibility and vascularity.13 Perhaps one of the most mineralized mass that arises following a clear history widely used applications of ultrasonography is imag- of trauma supports the diagnosis of myositis ossifi- ing-guided intervention.11 Ultrasonography can be cans.8 Phleboliths in a soft-tissue mass support the di- used to guide a minimally invasive technique to guide agnosis of a benign vascular lesion (Fig 2), whereas the the placement of the needle during biopsy into an area presence of fat suggests a lipomatous mass (Fig 3). of solid tissue. The utility of radiographs alone in the evaluation In a prospective study of 358 patients, ultrasonogra-

A B C D E F

Fig 1A–F. — Radiographic evaluation of osseous lesions in (A–D) aggressive and (E–F) benign bone tumors. (A) Anteroposterior view of the shoul- der in a patient aged 13 years with Ewing sarcoma. Shown is a permeative, lytic lesion with a wide zone of transition in the proximal humerus with a pathological fracture. Shown is a lamellate or layered periosteal reaction (black arrows), which is commonly seen in Ewing sarcoma and is suggestive of an aggressive lesion. (B) Frontal view of a femur in a patient aged 18 years with a dense, periosteal reaction in the femoral cortex with ossific matrix mineralization in the soft tissue of the thigh. These findings are consistent with osteosarcoma. (C) Frontal view of a tibia in a patient aged 13 years with an aggressive, sunburst periosteal reaction and an ossific sclerotic lesion in the proximal tibia consistent with osteosarcoma. (D) Frontal view of a femur in a patient aged 70 years with a lytic lesion with a wide zone of transition in the distal femur with stippled and curvilinear mineralized matrix (white arrowhead) suggestive of a cartilaginous lesion. Deeper cortical involvement and periosteal reaction (red arrow) were suggestive of a focally aggressive malignancy (chondrosarcoma). (E) Distal femur (white asterisk) compatible with nonossifying fibroma. (F) Lateral view of an elbow in a patient aged 18 years who was in pain. A dense, continuous solid periosteal reaction in the ulna was benign (red arrows) and had central lucency, the findings of which were compatible with osteoid osteoma. Subtle central calcification can be seen within the lucent lesion, reflecting calcification within the nidus, which was the inciting fibrovascular lesion.

162 Cancer Control April 2017, Vol. 24, No. 2 A B A B

C D C D E

Fig 2A–D. — Radiographic evaluation of soft-tissue masses. (A) Antero- Fig 3A–E. — Radiographic evaluation of soft-tissue masses. (A) Lateral posterior and (B) lateral views of the elbow in a man aged 24 years who was and (B) lateral hip imaging of the left thigh of a palpable, soft-tissue in pain and had a palpable mass that appeared to increase in size with phys- mass show a well-circumscribed area of lucency in the posterior thigh. ical exercise. Soft-tissue calcifications can be seen — many with lucent The presence of a lipomatous mass can be suggested based on the ra- centers — consistent with phleboliths and suggestive of a vascular mal- diographic findings. (C) Axial T1-weighted, gadolinium-enhanced, fat- formation (soft-tissue hemangioma). Note the lack of bony involvement or suppressed (D) T1-weighted and (E) sagittal T1-weighted imaging show joint-space abnormality, which can be excluded based on the radiographic the mass with a signal intensity that mimics subcutaneous fat and is findings. (C) T1- and (D) T2-weighted with fat-suppression magnetic reso- compatible with intramuscular lipoma. Note the uniform appearance of nance imaging confirmed the presence of the mass. Foci of low signal on the mass lacking nodules or thickened septations, findings suggestive of both sequences represent the phleboliths. The high T1-signal areas reflect benign intramuscular lipoma. the fatty overgrowth commonly seen in vascular malformations. phy was found to be an effective tool for triaging patients A B E with soft-tissue masses.14 Nearly 80% of lesions were characterized as benign based on ultrasonographic features; on further follow-up (clinical, MRI, or both), none of these patients had a malignant tumor.14 Of those lesions evaluated by MRI for suspicious or indeterminate findings on ultrasonography (large mass, deep-seated, painful solid components, or vascularity), fewer than 2% were histologically malignant at surgery.14 By contrast, a C D small study of soft-tissue tumors showed that 23% of the 43 patients studied were incorrectly diagnosed following initial findings on ultrasonography.15 A delay in diagnosis was observed (≤ 6 months) in 7 patients (5 with malignant tumors), and the most common error was interpret- ing a solid tumor as hematoma.15 These studies highlight the utility and pitfalls of ultrasonography.14,15 Although Fig 4A–E. — Role of skeletal scintigraphy in osteosarcoma. (A) Antero- it is a useful tool in evaluating patients with soft-tissue posterior view of the tibia shows increased density of the tibia and an associated periosteal reaction. (B) Lateral radiography shows a destructive masses and for use in guiding biopsy, ultrasonography lesion in the posterior cortex with a soft-tissue mass. (C) T1-weighted is not routinely performed in clinical practice as an initial and sagittal post–contrast magnetic resonance imaging confirm the imaging study, nor does its use preclude the need for ad- marrow-replacing bone lesion and extraosseous soft-tissue mass. (E) Whole-body skeletal scintigraphy for staging purposes shows in- ditional cross-sectional imaging evaluation. tense tracer uptake in the tibia without other sites of distant metastasis Because of the inability of sound waves to pene- or skip lesions. Computed tomography of the chest was necessary to trate bone, little role exists for ultrasonography in the exclude pulmonary metastasis. evaluation of bone lesions. phy is a cost-effective, whole-body imaging modality Nuclear Medicine for the evaluation of multifocal or disseminated osse- Radionuclide skeletal scintigraphy uses technetium-la- ous disease (Fig 4).11,17 beled diphosphonates, which adsorb to the crystalline Fludeoxyglucose F 18 (FDG) is the most commonly structure of hydroxyapatite. These tracers accumulate utilized radionuclide for PET. FDG-PET is a functional in areas of osteoblast activity, thus indicating areas of imaging technique that assesses tissue metabolism us- bone perfusion and bone turnover.16 Skeletal scintigra- ing radioisotopes that undergo positron emission de-

April 2017, Vol. 24, No. 2 Cancer Control 163 cay.17 FDG is a glucose analogue transported into the enlargement. An increase in tumor size (≤ 20%) ob- cell; however, once it is phosphorylated, it does not served after radiotherapy but prior to surgery has not undergo further metabolism and becomes trapped.12 been associated with worse outcomes.24 PET/CT can Fluorine F 18 then decays, producing 2 coincident also be used to guide biopsy. Soft-tissue sarcomas are positrons that can be used to produce the image. often heterogenous on MRI, with areas of necrosis or Glucose metabolism can be quantified by measuring myxoid components.25 If these areas are sampled, then maximum standardized uptake value.12 A caveat to the tumor may be inappropriately downgraded. PET/CT using PET/CT in musculoskeletal applications is the can assist in targeting hypermetabolic areas that can considerable overlap of standardized uptake values in increase the diagnostic yield of sampling (Fig 6).1 How- benign and malignant soft-tissue and bone lesions.18 ever, the evidence for the use of PET for guidance dur- In addition, non-neoplastic conditions (eg, inflam- ing biopsy is based on a small number of studies, so matory processes, trauma) can also result in abnor- mal uptake, mimicking malignancy and reducing the specificity of PET/CT.4 In clinical practice, CT is con- A B currently performed with PET for localizing the anatomical findings found on functional imaging. PET/CT has been investigated for biopsy guidance, staging, and treatment response.19 Typically, MRI of the primary lesion is performed; therefore, PET/CT is used 19 for the detection of metastatic disease. The lung is the C predominant site of metastatic disease from soft-tissue sarcoma, and it is unclear how well PET can detect metastases not already identified by CT of the chest.19 In several studies on the initial staging of soft-tissue sarcoma in adults using PET/CT, fewer than 5% of patients were upstaged as a result of PET/CT, and few patients had the management of their disease changed as a result.19,20 In a study of pediatric patients, FDG PET/CT was equally effective in identifying primary tumors (100%) compared with conventional imaging (CT and MRI) and was superior in the identification of metastatic lymph nodes (25% vs 95%) and bone metastasis (57% vs 90%).21 PET/CT was also found to have diagnostic benefit in the detection of bone metastases in osteosarcoma and pediatric Ewing Fig 5 A–C. — Utility of PET for evaluating osteosarcoma. Findings on (A) coronal and (B) axial PET demonstrate a large, peripherally avid tumor sarcoma, but it was less sensitive compared with conven- for fludeoxyglucose F 18 in the right hemithorax in a patient with primary tional CT in the detection of lung nodules (Fig 5).22 This osteosarcoma resected more than 7 years ago. (C) PET detected an area of finding underscores the importance of CT of the chest avidity in the left rib (red arrow) corresponding to a spiculated, periosteal in the staging workup of soft-tissue and bone sarcomas. reaction.PET = positron emission tomography. The utility of PET/CT beyond diagnostic CT of the chest for the initial and subsequent follow up of sarcomas is unclear.19 However, it may have added value in certain A B sarcomas (rhabdomyosarcoma, clear cell, epithelioid) in which the pattern of spread includes nonpulmonary sites, such as bone, retroperitoneum, and lymph nodes. Although the role of FDG-PET/CT is unclear in the initial diagnosis and staging of soft-tissue sarcoma, it can be used to assess therapeutic response early during treatment. Post-treatment changes measured by the FDG avidity of the tumor are an indicator of the ef- Fig 6A–B. — Utility of PET for evaluating osteosarcoma. Findings on fectiveness of the therapy. Earlier biochemical changes (A) axial CT and (B) axial-fused PET/CT demonstrate a large, periph- in a tumor can enable clinicians to evaluate therapeu- erally avid tumor for FDG in the right hemithorax in the patient from Fig 5. Primary osteosarcoma was resected more than 7 years ago tic effectiveness earlier than conventional anatomical from the left femur. Note how FDG avidity follows the areas of ossific imaging.23 In particular, sarcomas are challenging be- matrix seen on CT, thus suggesting the more metabolic regions and cause an increase in size often does not indicate poor potential area for biopsy. The center of the mass is likely necrotic and non-FDG avid, so it would be of lower diagnostic yield if biopsy was 24 response to therapy. Often, preoperative radiother- performed. CT = computed tomography, FDG = fludeoxyglucose F 18, apy causes hemorrhage or necrosis leading to tumor PET = positron emission tomography.

164 Cancer Control April 2017, Vol. 24, No. 2 the routine use of PET to guide biopsy in real time is A B C not routinely performed.26 More specifically, novel radiopharmaceuticals for use with PET, such as proliferation markers (fluoro- deoxythymidine F 18), bone-seeking agents (sodium fluoride F 18), amino-acid tracers (methionine C 11, fluoroethyltyrosine F 18), or biomarkers of neoangio- genesis (galacto-RGD F 18), are not routinely used in clinical practice. However, they can potentially eluci- Fig 7A–C. — CT evaluation of cortical lesions. (A) Radiography of the date the underlying biological mechanisms of muscu- humerus and (B) coronal CT show a benign, smooth, continuous, solid, periosteal reaction along the medial cortex. (C) Axial CT demonstrates loskeletal tumors and, with future research, possibly a cortical-based lucency in the cortex compatible with osteoid osteoma. contribute additional information to the grading, treat- The benign periosteal reaction was produced by inflammation caused by ment monitoring, and post-therapy assessment.27 osteoid osteoma. CT = computed tomography. Applications for PET/MRI continue to grow in clinical practice.17 The literature on PET/MRI is not yet available for all tumors, but PET/MRI may be of benefit A B in staging and assessing treatment response in bone and soft-tissue sarcomas.17 In general, PET/MRI may be indicated in malignancies that require high soft-tissue contrast for visualization, and it will be of utility in the pediatric population because radiation exposure can be reduced.17,28

Computed Tomography C When the radiographic features of a bone lesion are indeterminate or aggressive, cross-sectional imaging is usually obtained. Because of its superior ability to visualize bony detail and evaluate cortical bone, CT is often the modality used to evaluate tumors located within the periosteal or cortical regions (eg, osteoid osteoma; Fig 7).5 CT can better demonstrate subtle mineralization or calcification that may not be appreciated on ra- Fig 8A–C. — CT evaluation of the small bones in the feet. (A) Oblique diography.4 Evaluation of flat bones and small bones view of the foot shows a lucent lesion with a sclerotic center confirmed by (B) long-axis and (C) axial CT. CT demonstrates a lucent of the hands and feet are also better evaluated with CT lesion consistent with osteoid osteoma as well as the calcified fibro- (Fig 8).4 Due to the complex anatomy in the spine and vascular nidus, which is the inciting lesion. CT also shows the dis- pelvis, CT may also be a preferred modality or used in ruption of the cortical bone that was not observed on radiography. CT = computed tomography. adjunct to MRI.4 CT may also be the most appropriate imaging modality for patients who are obese, patients with pacemakers, and when MRI is not feasible.1 A B C Earlier literature comparing MRI and CT for the evaluation of bone tumors showed that MRI was superior for initial staging.4,29-31 Zimmer et al29 and Ho- geboom et al30 studied features of bone tumors such as cortical destruction and involvement of marrow, soft tissue, joints, and neurovascular structure on CT and MRI.4 Both studies showed superiority of CT for evaluating cortical bone involvement, but MRI better demonstrated neuromuscular, and joint and compartmental involvement, as well as intramedullary extent.29,30 Both studies also favored MRI if both CT and MRI were available.29,30 One group demonstrated no statistical difference between CT and MRI in the evaluation of the extent of tumor involvement in 183 patients with Fig 9A–C. — (A) Lateral radiography of the tibia. (B) Sagittal T1-weight- primary bone tumors (Fig 9).1,4,31 ed MRI. (C) Sagittal-reconstruction CT in a boy aged 15 years with left tibial adamantinoma. Radiographs, CT, and MRI all show the eccentric, CT is useful in evaluating mineralized matrix and cortical-based location and extent of the lesion with an intact cortex. bone changes related to soft-tissue sarcoma.32 It can CT = computed tomography, MRI = magnetic resonance imaging.

April 2017, Vol. 24, No. 2 Cancer Control 165 be helpful in distinguishing soft-tissue sarcoma from Myxomatous stroma demonstrates mild enhance- myositis ossificans — a distinction that can be diffi- ment, which is in contrast to cystic areas or necrosis cult on MRI, as both can have nonspecific features on that does not enhance. T1- and T2-weighted imaging — because CT can de- Oftentimes, the degree of enhancement is visu- lineate the zonal pattern of mineralization character- ally equivocal. In these cases, subtraction sequences istic of myositis ossificans.32 can be used. Precontrast, T1-weighted images are subtracted from the postcontrast T1-weighted se- Magnetic Resonance Imaging quences by software to yield areas of true enhance- Evaluation of bone tumors using MRI can help clini- ment (Fig 12). The utility of gadolinium in imaging cians identify the extraosseous component, if pres- bone tumors is controversial; nevertheless, the pat- ent, and the compartmental location can be well tern and degree of contrast enhancement can be evaluated — both are important factors in the preop- useful, particularly when evaluating for extraosse- erative evaluation. In addition, a diagnosis other than ous extension. a tumor (eg, radiographically occult fracture, osteone- Important Features: The location, size, and sig- crosis) may be elucidated on MRI.3,5 It is important to nal features of a lesion should be assessed. In general, remember that radiography plays an important and heterogeneous signal is seen in sarcomas, including complementary role in the evaluation of bone tumors. areas of mixed tissue, necrosis, or hemorrhage. If ar- Most bone lesions (malignant and benign) have high eas of intralesional fat or calcification are observed, T1 and T2 relaxation times, producing nonspecific or the mass is arising from a nerve, then the differ- low T1 and high T2 signals.2 This fact makes the ren- ential diagnosis can be appropriately narrowed as a dering of a particular histological diagnosis all the lipomatous or nerve sheath tumor, respectively. The more difficult when evaluated in the absence of radiography. In-phase and out-of-phase (also known as A B opposed) imaging can be useful in the evaluation of bone lesions, particularly in delineating red marrow hyperplasia from metastatic disease. In general, MRI has replaced CT as the imaging modality of choice in the evaluation of soft-tissue masses.1 Based on the findings of 133 soft-tissue tumors, one group found that MRI and contrast-enhanced CT were comparable when determining tumor size and involvement of surrounding structures.1,31 However, in clinical practice, MRI is the preferred imaging modality for the detection, evaluation, assessment, staging, and follow up of soft-tissue tumors. Fig 10A–B. — MRI detection of lymphadenopathy. (A) Axial, post–contrast MRI of the thigh demonstrates a large, posterior, compartment The advantages of MRI include establishing a dif- mass with peripheral enhancement consistent with leiomyosarcoma ferential diagnosis of the lesion, precise compartmen- proven by biopsy. (B) Axial MRI of the ipsilateral inguinal region dem- tal localization, and assessment of neuromuscular onstrates an enlarged, enhancing lymph node compatible with metastatic disease. With the exception of certain histological subtypes, and joint involvement. In addition, MRI can be used nodal metastatic disease is not commonly seen in soft-tissue sarcomas. to visualize the entire compartment, which can dem- MRI = magnetic resonance imaging. onstrate metastatic lymphadenopathy — seen only in certain sarcomas (Fig 10). MRI is excellent at delineating soft-tissue lesions, and a specific diagnosis can A B C be obtained in many instances by evaluating lesion signal intensity, location, growth pattern, and other unique intrinsic properties.33 Unless a specific diagnosis can be made, the lesion should be considered indeterminate and then biopsied in close consultation with an orthopedic oncologist. Unless it is contraindicated, contrast should be administered when evaluating soft-tissue tumors. Fig 11A–C. — Heterogenous soft-tissue sarcoma. (A) Axial T1-weighted, (B) T2-weighted, and (C) post–contrast MRI of a thigh with malignant Contrast enhancement delineates areas of necrosis myxoid liposarcoma. High T2-signal areas throughout the mass repre- within the lesion and allows differentiation of myxoid sent myxoid stroma, a gelatinous material comprised of glycosamino- material from fluid foci (Fig 11). Myxoid tissue is a ge- glycans, commonly seen in soft-tissue sarcoma. Myxomatous stroma demonstrates mild enhancement; this is in contrast to cystic areas or latinous material comprised of glycosaminoglycans, necrosis, neither of which enhance on MRI. and it is commonly seen in soft-tissue sarcomas.34 MRI = magnetic resonance imaging.

166 Cancer Control April 2017, Vol. 24, No. 2 presence of areas of low signal on multiple sequences more difficult to detect on MRI due to the low signal can suggest the presence of fibrous tissue. of cortical bone on all sequences; in these cases, CT The high soft-tissue contrast of MRI enables the can be used as an adjunct imaging modality. clinician to assess the relationship between the tu- Typically, a rim of low signal surrounding the tumor and neurovascular bundle (Fig 13). If the contact mor can be seen in sarcomas. Malignant tumors push between the tumor and the adjacent neurovascular away normal tissue around them as they enlarge; structures exceeds 180 degrees, then encasement therefore, sarcomas do not infiltrate the anatomical should be suspected.35 compartments and fascial borders until late in their Direct osseous involvement by soft-tissue sarco- course. Local growth of soft-tissue sarcoma occurs ma rarely occurs, but the clinician should still eval- in a radial fashion, compressing surrounding fibrous uate for it because its presence has been shown to connective tissue, which, in combination with associ- correlate with disease-related mortality.32,36 Osseous ated inflammatory reaction, forms a “pseudocapsule” invasion of the medullary cavity can be assessed on around the tumor (Fig 14).33,37 MRI by signal changes in the normal fat-containing Another important feature to assess is the pres- marrow (see Fig 13). Cortical involvement may be ence of peritumoral high T2 or fluid signal changes that can be seen in benign and malignant tumors.37 A B C This may be due to the increased water content in the tissues, known as peritumoral edema (see Fig 14).38 Typically, areas of peritumoral edema enhance, and they have been attributed to microscopic tumors, inflammatory reactions, vascular congestion and hy- perperfusion, or edema.38 In a study by White et al,38 malignant cells were histologically identified in tissue D E beyond the tumor margin in 10 out of 15 cases. Iden- tification of malignant cells beyond the margin of the tumor necessitates treatment modifications, such as extension of the radiation field or repeated resection to reduce the risk of local recurrence. In addition to the initial staging of primary soft- tissue and bone tumors, MRI is the preferred modal- Fig 12A–E. — Subtraction MRI in a woman aged 76 years with undifferenti- ity for the evaluation of locally recurrent disease.17,37 ated pleomorphic sarcoma. Shown are (A) precontrast, T1-weighted without Surgical and radiation changes can produce ab- fat suppression, (B) T1-weighted with fat suppression, (C) T2-weighted, (D) gadolinium-enhanced, and (E) subtraction MRI. High T1-signal areas normal signals related to fibrosis, edema, and me- could represent blood products or a solid tumor. Diffuse gadolinium enhance- tallic susceptibility artifact, thus making imaging ment can be seen throughout the lesion — a finding that does not aid in differentiation. Subtraction MRI (obtained from precontrast subtracted from postcontrast) reveals only areas of true enhancement in the periphery of the mass. A B Subtraction MRI can be useful in hemorrhagic tumors and in cases of recurrent disease, where true enhancement can be difficult to differentiate from blood products or surrounding postoperative or therapy-related changes. MRI = magnetic resonance imaging.

A B

Fig 14A–B. — Pseudocapsule and peritumoral edema. (A) Sagittal, T2- Fig 13A–B. — Neurovascular encasement and bony involvement shown weighted, fat-suppressed, and (B) postcontrast magnetic resonance in axial, fat-suppressed, postcontrast MRI. (A) The anterior tibial neuro- imaging of the thigh in a patient with undifferentiated pleomorphic sar- vascular bundle (white arrow) is completely encased by the mass. coma. (A) The dark signal rim surrounding the mass (white arrowheads) (B) Invasion of the adjacent fibula is demonstrated by the soft-tissue is caused by the centrifugal growth of the tumor pushing and compress- mass (white arrow). Both findings are of utmost importance to evaluate ing the adjacent normal tissue. (B) Long “tails” of feathery high signal and report on MRI, because they are critical to appropriate surgical plan- proximal and distal to the mass (A; red arrows) demonstrate enhance- ning. MRI = magnetic resonance imaging. ment (red arrows).

April 2017, Vol. 24, No. 2 Cancer Control 167 interpretation complex.39 Postoperative fluid col- T1 signal, thus mimicking a marrow-replacing le- lections or seromas can complicate the detection of sion. Red marrow also contains fat — albeit less when recurrent disease, which most often presents with compared with yellow marrow — and, therefore, can enhancing nodules in the tumor bed (Fig 15).39,40 be differentiated from true lesions by chemical shift However, enhancing mural nodules within seromas MRI. A technique described by Costa et al43 involves may not necessarily represent recurrent disease and drawing regions of interest of equal size over the ab- should be followed up.41 Another common finding normal areas on both the in-phase and out-of-phase is treatment-related changes of the bone marrow, imaging. The signal intensity ratio of the marrow on which can be difficult to differentiate from tumor. the out-of-phase and in-phase imaging is calculated. T1-weighted sequences and chemical-shift MRI can A signal intensity ratio below 0.80 is typical of a non- be useful tools for delineating changes in bone mar- neoplastic process.43 A study by Zajick et al46 showed row from neoplasms.42 that a 20% decrease in signal intensity on the out-of- phase images from the in-phase images was a reli- Novel Techniques able feature for distinguishing benign from malignant Chemical Shift and In- and Out-of-Phase Magnetic bone marrow in the spine. Although using a signal in- Resonance Imaging tensity ratio higher than 0.80 has been proposed to In-phase and out-of-phase (opposed) MRI allows the identify neoplastic lesions, chemical shift MRI has not clinician to detect the presence of fat in lesions, so been consistently shown to differentiate malignant this imaging modality may be useful for delineating from benign tumors.46 whether a signal abnormality seen in the marrow is likely caused by red marrow hyperplasia or a mar- Diffusion-Weighted Magnetic Resonance Imaging row-replacing process. Normal yellow marrow has a Diffusion-weighted is a noncontrast, functional MRI high fat content with a hyperintense signal on non– technique that has been investigated for the char- fat-suppressed T1 sequences. Normal bone marrow is acterization of tumors throughout the body.47 The rich in both fat and water, but their relative amounts signal intensity of diffusion-weighted MRI relies on is the main factor affecting signal intensity of marrow the Brownian motion, or the microscopic motion of on MRI.43,44 When both fat and water are present in water molecules within tissues.48 The apparent dif- the same voxel, this combination results in the sup- fusion coefficient (ADC) is a quantitative measure of pression of signal intensity on out-of-phase imag- Brownian motion. In highly cellular environments, ing.45 Marrow-replacing processes result in a lower free diffusion of water restricted, resulting in low T1 signal and lack of suppression of signal intensity ADCs. Acellular tissue (ie, necrosis) allows free dif- on out-of-phase imaging (Fig 16).45 Hematopoietic fusion in all directions, resulting in high ADCs.49 or red marrow can produce focal areas of decreased Areas of restricted diffusion (ie, because of high cel-

Fig 16. — Chemical shift MRI. Coronal postcontrast MRI shows a meta- Fig 15. — MRI detection of recurrent disease. Axial, postcontrast MRI static lesion in the humerus. In- and opposed-phase axial imaging show in a patient with resected, high-grade sarcoma undergoing surveillance. no signal decrease, indicating that no fat-containing voxels can be found Areas of diffuse enhancement (white arrowheads) are consistent with in the lesion, thereby suggesting this patient has a malignant lesion. postoperative change and scarring. However, 2 solid, enhancing nod- MRI = magnetic resonance imaging. ules adjacent to the humeral head (red arrows) reflect recurrent disease. Images courtesy of Laura M. Fayad, MD, Johns Hopkins Medicine, MRI = magnetic resonance imaging. Baltimore, Maryland.

168 Cancer Control April 2017, Vol. 24, No. 2 lularity) have higher signal intensity on diffusion- Proton Magnetic Resonance Spectroscopy weighted MRI but lower signal intensity on ADC maps Proton magnetic resonance spectroscopy is a technique (Fig 17).12,43 The ADC value for a specific region of in- used to characterize tumors based on their molec- terest is calculated by plotting the change in signal of ular properties; similar to diffusion-weighted MRI, the region as it varies with different diffusion gradient it can be performed without intravenous contrast.54 strengths (b values).50 The metabolic “footprint” of a lesion is determined Diffusion-weighted MRI has been used to better based on the signal of water, lipids, and various me- evaluate treatment response in patients with sarcoma tabolite content.54 Choline-containing compounds receiving chemotherapy.43,49 However, the literature are constituents of cell membrane phospholip- shows variable results in the ability of diffusion-weight- ids and are a reflection of cell-membrane turnover. ed MRI to distinguish between benign and malignant Therefore, the presence of a spectroscopic choline soft-tissue masses.49,51 A study by Subhawong et al49 de- peak is suggestive of malignancy.54-56 The presence termined that a threshold mean ADC of 2.5 provided a of a choline peak is supportive of malignancy, al- 100% specificity rate for predicting a cyst and ruling out though some metabolically active benign tumors a solid tumor. Demehri et al52 demonstrated the ability and abscesses may also have a choline peak. of a minimum ADC to differentiate benign from malig- Magnetic resonance spectroscopy is not widely nant peripheral nerve sheath tumors. used in clinical practice for musculoskeletal applica- Few studies have examined the utility of diffusion- tions, but it shows promise in the noncontrast magnet- weighted MRI in the evaluation of bone lesions; how- ic resonance evaluation of bone and soft-tissue lesions. ever, a study by Ahlawat et al47 determined the minimum ADC threshold of 0.9 × 10−3 mm2/second and Habitat Imaging 1.4 × 10−3 mm2/second (mean ADC) for differentiating Several groups have been working on the quantitative benign and malignant histology (see Fig 17). A study evaluation of tumor subregions in the setting of soft- of patients with osteosarcoma suggests that change in tissue sarcoma.57-61 Utilizing several computer-aided ADC may be useful in assessing response to chemo- techniques, including clustering algorithms and tex- therapy and can differentiate areas of granulation tis- ture-feature analyses, one group noninvasively evalu- sue and scarring from viable cellular tumor.50,53 ated tumor subregions, or “habitats,” to better elucidate Although it is not routinely used, diffusion- the heterogeneity of soft-tissue sarcoma. T1-weighted, weighted MRI is an unenhanced method that may gadolinium-enhanced, and fluid-sensitive pretreatment provide helpful clinical information, and it can be MRI was analyzed in the extremities of patients with used in situations in which intravenous contrast can- soft-tissue sarcoma.57 Segmentation based on the pixel not be administered due to contraindications or pa- signal intensity was performed, followed by an analysis tient refusal. within each distinctive subregion to predict metastatic disease and histological necrosis.57 From this analysis, A B habitat maps were derived based on combinations of pixel intensity combinations from the 2 sequences, demonstrating distinct, spatial intratumoral subregions (Fig 18). The yellow-color habitat map corresponds to a high T1, postcontrast/low, T2 signal (inferred as high vascularity and cellularity), which was both predictive and prognostic. The yellow subregion correlated with decreased overall survival (P = .03) and, if present in more than 18% of the tumor slice, is predictive for the development of metastatic or locally recurrent disease. Based on texture-feature analysis, pretreatment MRI predicted that the rate of metastatic development in patients with soft-tissue sarcoma was 72.4%.61 Continued investigation into the change in habitats on pre- and post-treatment MRI is also being conducted.62 Habitat imaging is an evolving area of study that may be useful Fig 17A–B. — Diffusion-weighted magnetic resonance imaging in the evaluation of the lesion in Fig 16. The lesion shows high signal at (A) different for more accurately assessing prognosis and tailoring b values and (B) low signal on the apparent diffusion coefficients map, personalized therapies.57 findings indicative of restricted diffusion. Restricted diffusion is caused by the restricted motion of water in a cellular environment and suggests that a lesion is malignant. Conclusions Images courtesy of Laura M. Fayad, MD, Johns Hopkins Medicine, Progress has been made in the imaging evaluation of Baltimore, Maryland. musculoskeletal tumors. Detecting lesions on mag-

April 2017, Vol. 24, No. 2 Cancer Control 169 masses: a current perspective. AJR Am J Roentgenol. 2000;175(3):575-587. A Pre treatment B C 10. Manaster BJ. Soft-tissue masses: optimal imaging protocol and reporting. AJR Am J Roentgenol. 2013;201(3):505-514. 11. Balach T, Stacy GS, Haydon RC. The clinical evaluation of soft tissue tumors. Radiol Clin North Am. 2011;49(6):1185-1196, vi. 12. Afonso PD, Mascarenhas VV. Imaging techniques for the diagnosis of soft tissue tumors. Rep Med Imag. 2015;2015(8):63-70. 13. Carra BJ, Bui-Mansfield LT, O’Brien SD, et al. Sonography of musculoskeletal soft-tissue masses: techniques, pearls, and pitfalls. AJR Am J Roentgenol. 2014;202(6):1281-1290. 14. Lakkaraju A, Sinha R, Garikipati R, et al. Ultrasound for initial evaluation and triage of clinically suspicious soft-tissue masses. Clin Radiol. D Post treatment E F 2009;64(6):615-621. 15. Doyle AJ, Miller MV, French JG. Ultrasound of soft-tissue masses: pitfalls in interpretation. Australas Radiol. 2000;44(3):275-280. 16. Wong KK, Piert M. Dynamic bone imaging with 99mTc-labeled diphosphonates and 18F-NaF: mechanisms and applications. J Nucl Med. 2013;54(4):590-599. 17. Ilaslan H, Schils J, Nageotte W, et al. Clinical presentation and imaging of bone and soft-tissue sarcomas. Cleve Clin J Med. 2010;77(suppl 1): S2-S7. 18. Shin DS, Shon OJ, Han DS, et al. The clinical efficacy of (18)F-FDG- PET/CT in benign and malignant musculoskeletal tumors. Ann Nucl Med. Fig 18A–F. — Habitat imaging in a patient aged 48 years with high-grade, 2008;22(7):603-609. undifferentiated sarcoma in the left thigh. (A, D) Axial, contrast-enhanced, 19. Roberge D, Vakilian S, Alabed YZ, et al. FDG PET/CT in initial staging (B, E) T2-weighted, fat-suppressed, and (C, F) habitat color mapping ob- of adult soft-tissue sarcoma. Sarcoma. 2012;2012:960194. 20. Fuglo HM, Jorgensen SM, Loft A, et al. The diagnostic and prognostic tained following computed tumor segmentation. Panels A to C show pre- value of (1)(8)F-FDG PET/CT in the initial assessment of high-grade bone treatment MRI. Panels D to F show post-treatment MRI. The color maps and soft tissue sarcoma. A retrospective study of 89 patients. Eur J Nucl illustrate distinct, intratumoral subregions, or habitats, derived from pixel Med Mol Imaging. 2012;39(9):1416-1424. signal intensities. The degree of spatial heterogeneity within the tumor is 21. Volker T, Denecke T, Steffen I, et al. Positron emission tomography better delineated on the habitat color map. Note the change in distribution for staging of pediatric sarcoma patients: results of a prospective multi- and size of the habitats between pre- and post-contrast color maps. The center trial. J Clin Oncol. 2007;25(34):5435-5441. yellow subregion comprises more than 18% of the tumor pretreatment 22. Quartuccio N, Fox J, Kuk D, et al. Pediatric bone sarcoma: diagnostic color map in panel C, thus inferring a poor prognosis. This patient had performance of (1)(8)F-FDG PET/CT versus conventional imaging for initial staging and follow-up. AJR Am J Roentgenol. 2015;204(1):153-160. metastatic disease to the lungs and died less than 1 year after diagnosis. 23. Sheikhbahaei S, Marcus C, Hafezi-Nejad N, et al. Value of FDG MRI = magnetic resonance imaging. PET/CT in patient management and outcome of skeletal and soft tissue sarcomas. PET Clin. 2015;10(3):375-393. 24. Delisca GO, Alamanda VK, Archer KR, et al. Tumor size increase following preoperative radiation of soft tissue sarcomas does not affect netic resonance imaging has improved the ability of cli- prognosis. J Surg Oncol. 2013;107(7):723-727. 25. Blockx PP, Carp L. Nuclear medicine imaging. In: De Schepper AM, nicians to differentiate normal anatomical variants from Parizel PM, Beuckeleer LD, et al. Imaging of Soft Tissue Tumors. 2nd ed. true lesions and to characterize these lesions; additional New York: Springer; 2001:45-69. 26. Becher S, Oskouei S. PET imaging in sarcoma. Orthop Clin North advances and the development of novel techniques will Am. 2015;46(3):409-415, xi. continue to enhance this ability. Other conventional im- 27. Wieder HA, Pomykala KL, Benz MR, et al. PET tracers in musculoskeletal disease beyond FDG. Semin Musculoskel Radiol. 2014;18(2):123-132. aging modalities continue to play an integral and com- 28. Buchbender C, Heusner TA, Lauenstein TC, et al. Oncologic PET/ plementary role in the evaluation of musculoskeletal MRI, part 2: bone tumors, soft-tissue tumors, melanoma, and lymphoma. J Nucl Med. 2012;53(8):1244-1252. tumors. Habitat imaging that uses magnetic resonance 29. Zimmer WD, Berquist TH, McLeod RA, et al. Bone tumors: mag- imaging can identify and quantify tumor heterogeneity netic resonance imaging versus computed tomography. Radiology. 1985;155(3):709-718. and provide health care professionals with useful clini- 30. Hogeboom WR, Hoekstra HJ, Mooyaart EL, al. MRI or CT in the pre- cal and prognostic information that can shape person- operative diagnosis of bone tumours. Eur J Surg Oncol. 1992;18(1):67-72. 31. Panicek DM, Gatsonis C, Rosenthal DI, et al. CT and MR imaging in alized and adaptive therapeutic regimens. the local staging of primary malignant musculoskeletal neoplasms: report of the Radiology Diagnostic Oncology Group. Radiology. 1997;202(1):237-246. References 32. Robinson E, Bleakney RR, Ferguson PC, et al. Oncodiagnosis panel: 2007: multidisciplinary management of soft-tissue sarcoma. Radiographics. 1. American College of Radiology. ACR appropriateness criteria: soft- 2008;28(7):2069-2086. tissue masses. Published 1995. Last reviewed 2012. https://acsearch.acr. 33. Walker EA, Salesky JS, Fenton ME, et al. Magnetic resonance org/docs/69434/Narrative. Accessed March 6, 2017. imaging of malignant soft tissue neoplasms in the adult. Radiol Clin North Am. 2. Sundaram M, McLeod RA. MR imaging of tumor and tumorlike 2011;49(6):1219-1234, vi. lesions of bone and soft tissue. AJR Am J Roentgenol. 1990;155(4):817-824. 34. Baheti AD, Tirumani SH, Rosenthal MH, et al. Myxoid soft-tissue 3. Stacy GS, Peabody TD. Staging of bone tumors: a review with illus- neoplasms: comprehensive update of the taxonomy and MRI features. trative examples. AJR Am J Roentgenol. 2006;186(4):967-976. AJR Am J Roentgenol. 2015;204(2):374-385. 4. American College of Radiology. ACR appropriateness criteria: 35. Holzapfel K, Regler J, Baum T, et al. Local staging of soft-tissue primary bone tumors. Published 1995. Last reviewed 2013. https:// sarcoma: emphasis on assessment of neurovascular encasement-value acsearch.acr.org/docs/69421/Narrative. Accessed March 6, 2017. of MR imaging in 174 confirmed cases. Radiology. 2015;275(2):501-509. 5. Balach T, Stacy GS, Peabody TD. The clinical evaluation of bone 36. Panicek DM, Go SD, Healey JH, et al. Soft-tissue sarcoma involving tumors. Radiol Clin North Am. 2011;49(6):1079-1093, v. bone or neurovascular structures: MR imaging prognostic factors. Radiology. 6. Miller TT. Bone tumors and tumorlike conditions: analysis with con- 1997;205(3):871-875. ventional radiography. Radiology. 2008;246(3):662-674. 37. Zhang Z, Deng L, Ding L, et al. MR imaging differentiation of malignant 7. American College of Radiology. ACR–SPR–SSR practice param- soft tissue tumors from peripheral schwannomas with large size and hetero- eter for the performance and interpretation of magnetic resonance imaging geneous signal intensity. Eur J Radiol. 2015;84(5):940-946. (MRI) of bone and soft tissue tumors. Updated 2015. www.acr.org/~/media/ 38. White LM, Wunder JS, Bell RS, et al. Histologic assessment of ACR/Documents/PGTS/guidelines/MRI_Bone_Soft_Tissue.pdf. Accessed peritumoral edema in soft tissue sarcoma. Int J Radiat Oncol Biol Physics. March 6, 2017. 2005;61(5):1439-1445. 8. Wu JS, Hochman MG. Soft-tissue tumors and tumorlike lesions: a 39. Bancroft LW. Postoperative tumor imaging. Semin Musculoskel Radiol. systematic imaging approach. Radiology. 2009;253(2):297-316. 2011;15(4):425-438. 9. Kransdorf MJ, Murphey MD. Radiologic evaluation of soft-tissue 40. Kransdorf MJ, Murphey MD. Soft tissue tumors: post-treatment im-

170 Cancer Control April 2017, Vol. 24, No. 2 aging. Radiol Clin North Am. 2006;44(3):463-472. 41. Lantos JE, Hwang S, Panicek DM. Benign mural nodules within fluid collections at MRI after soft-tissue sarcoma resection. AJR Am J Roentgenol. 2014;202(6):1297-1302. 42. Disler DG, McCauley TR, Ratner LM, et al. In-phase and out-of- phase MR imaging of bone marrow: prediction of neoplasia based on the detection of coexistent fat and water. AJR Am J Roentgenol. 1997;169(5):1439-1447. 43. Costa FM, Canella C, Gasparetto E. Advanced magnetic resonance imaging techniques in the evaluation of musculoskeletal tumors. Radiol Clin North Am. 2011;49(6):1325-1358, vii-viii. 44. Hwang S, Panicek DM. Magnetic resonance imaging of bone marrow in oncology, part 1. Skel Radiol. 2007;36(10):913-920. 45. Ragab Y, Emad Y, Gheita T, et al. Differentiation of osteoporotic and neoplastic vertebral fractures by chemical shift {in-phase and out-of phase} MR imaging. Eur J Radiol. 2009 Oct;72(1):125-133. 46. Zajick DC Jr, Morrison WB, Schweitzer ME, et al. Benign and malignant processes: normal values and differentiation with chemical shift MR imaging in vertebral marrow. Radiology. 2005;237(2):590-596. 47. Ahlawat S, Khandheria P, Subhawong TK, et al. Differentiation of benign and malignant skeletal lesions with quantitative diffusion weighted MRI at 3T. Eur J Radiol. 2015;84(6):1091-1097. 48. Khoo MM, Tyler PA, Saifuddin A, et al. Diffusion-weighted imaging (DWI) in musculoskeletal MRI: a critical review. Skelet Radiol. 2011;40(6):665-681. 49. Subhawong TK, Durand DJ, Thawait GK, et al. Characterization of soft tissue masses: can quantitative diffusion weighted imaging reliably distinguish cysts from solid masses? Skelet Radiol. 2013;42(11):1583-1592. 50. Subhawong TK, Jacobs MA, Fayad LM. Diffusion-weighted MR Imaging for Characterizing Musculoskeletal Lesions. Radiographics. 2014;34(5):1163-1177. 51. Einarsdottir H, Karlsson M, Wejde J, et al. Diffusion-weighted MRI of soft tissue tumours. Eur Radiol. 2004;14(6):959-963. 52. Demehri S, Belzberg A, Blakeley J, Fayad LM. Conventional and functional MR imaging of peripheral nerve sheath tumors: initial experience. AJNR Am J Neuroradiol. 2014;35(8):1615-1620. 53. Hayashida Y, Yakushiji T, Awai K, et al. Monitoring therapeutic responses of primary bone tumors by diffusion-weighted image: initial results. Eur Radiol. 2006;16(12):2637-2643. 54. Fayad LM, Jacobs MA, Wang X, et al. Musculoskeletal tumors: how to use anatomic, functional, and metabolic MR techniques. Radiology. 2012;265(2):340-356. 55. Doganay S, Altinok T, Alkan A, et al. The role of MRS in the differentiation of benign and malignant soft tissue and bone tumors. Eur J Radiol. 2011;79(2):e33-e37. 56. Fayad LM, Bluemke DA, McCarthy EF, et al. Musculoskeletal tumors: use of proton MR spectroscopic imaging for characterization. J Magn Res Imaging. 2006;23(1):23-28. 57. Raghavan M, Farhidzadeh H, Hall LO. MRI defined ecologic habits in extremity soft tissue sarcomas: characterization and quantification of tumor heterogeneity and potential implications of patient outcomes-early experience. Presented at: International Skeletal Society; Maui, HI; December 1, 2015. Abstract SSJ15-03. 58. Farhidzadeh H, Zhou M, Goldgof DB, et al. Prediction of treatment response and metastatic disease in soft tissue sarcoma. In: Aylward S, Hadjiiski LM, eds. Medical Imaging 2014: Computer-Aided Diagnosis. Vol. 9035. San Diego, CA: SPIE; 2014. 59. Farhidzadeh H, Chaudhury B, Zhou M, et al. Prediction of treatment outcome in soft tissue sarcoma based on radiologically defined habitats. In: Hadjiiski LM, Tourassi GD, eds. Medical Imaging 2015: Computer-Aided Diagnosis. Vol. 9414. San Diego, CA: SPIE; 2015. 60. Farhidzadeh H, Goldgof DB, Hall LO, et al. Texture feature analysis to predict metastatic and necrotic soft tissue sarcomas. Presented at: IEEE International Conference on Systems, Man, and Cybernetics; Hong Kong; October 9–12, 2015. 61. Farhidzadeh H, Chaudhury B, Scott JG, et al. Signal intensity analysis of ecological defined habitat in soft tissue sarcomas to predict metastasis development. In: Tourassi GD, Armato SG. Medical Imaging 2016: Computer-Aided Diagnosis. San Diego, CA: SPIE; 2016. 62. Farhidzadeh H, Goldgof DB, Hall LO, et al. A quantitative histogram- based approach to predict treatment outcome for soft tissue sarcomas using pre- and post-treatment MRIs. Presented at: IEEE International Conference on Systems, Man, and Cybernetics; Budapest, Turkey; October 9–12, 2016.

April 2017, Vol. 24, No. 2 Cancer Control 171