Acute Spinal Subdural Hematoma: .MR and CT Findings with Pathologic Correlates M. Judith Donovan Post, Jose L. Becerra, Parley W. Madsen, William Puckett, Robert M. Quencer, R. P. Bunge, and Evelyn M. L. Sklar PURPOSE: To determine the MR and CT findings that characterize acute spinal subdural hema­ toma (ASSH) . METHODS: The MR , CT, and clinical findings in three patients with surgically proved ASSH were reviewed and also correlated with the postmortem MR , CT, and cryomicrotome findings in three other patients, two with ASSH and one with an acute spinal epidural hematoma. RESULTS: Imaging findings in ASSH included: (a) hyperdense lesions on plain CT within the dural sac, distinct from the adjacent low-density epidural fat and silhouetted against the lower-density spinal cord and cauda equina, which it compressed; (b) lack of direct continuity with the adjacent osseous structures; (c) clumping, loculation, and streaking of blood within the dural sac on both MR and CT; and (d) an inhomogeneous and variable signal intensity to the ASSH on all MR pulse sequences, but, nevertheless, a striking low signal intensity on T2-weighted spin-echo or T2- weighted gradient-echo to a major part of the ASSH because of deoxyhemoglobin. Plain CT was most helpful in compartmentalizing the hematoma. CONCLUSION: When MR and plain CT are obtained as complementary studies, they provide characteristic findings that allow the prompt diagnosis of ASSH. Index terms: Hematoma, spinal; Hematoma, subdural; Spine, computed tomography; Spine, magnetic resonance; Pediatric neuroradiology AJNR Am J Neuroradiol 15:1895-1905, Nov 1994 Acute spinal subdural hematoma (ASSH) is this condition can affect patient outcome favor­ an uncommonly recognized condition that if left ably. Despite the importance of early diagnosis, untreated can result in severe irreversible neu­ the imaging characteristics of ASSH, unlike rologic deficits (1-7). Like acute spinal epidural acute spinal epidural hematoma ( 11-29), have hematomas, these ASSHs can expand rapidly not been well elucidated. Recent experience at and extend over many spinal segments, caus­ our institution with three patients with surgically ing sudden spinal cord and/or cauda equina proved ASSH prompted us to review their clin­ compression ( 1-19). Because rapid surgical ical and imaging findings and correlate them decompression can reverse neurologic deficits with the postmortem magnetic resonance (MR) or halt their progression, prompt recognition of and computed tomographic (CT) scans and cryomicrotome data in three other patients with spinal hematomas to determine what imaging Received November 19, 1993; accepted after revision March 9, 1994. patterns are characteristic of this condition. Presented at the 31st Annual Meeting of the American Society of Neu­ roradiology, Vancouver, Canada, May 1993. Supported by grant I POl NS-28059-01 A 1 from the National Institute of Neurological Disorders and Subjects and Methods Stroke and by the Miami Project to Cure Paralysis. From the Departments of Radiology (Neuroradiology Section) Postmortem material was obtained from three children (M.J.D.P. , J .L.B., R.M.Q. , E.M.L.S.) and Neurological Surgery (P.W.M.) who had sustained significant trauma from motor vehicle and the Miami Project to Cure Paralysis (W.P. , R.P.B. ), University of Miami accidents or shaking episodes. They were either dead on School of Medicine/ Jackson Memorial Medical Center, Fl a. arrival or died within 2 days of the injuries. Their vertebral Address reprint requests toM. Judith Donovan Post, MD, Neuroradiol­ columns with the spinal cords in situ were removed at ogy Section, MRI Center, 1115 NW 14th St, Miami, FL 331 36. autopsy and were placed immediately in 10% buffered AJNR 15:1895-1905, Nov 1994 0195-6108/ 94/1510-1895 formalin for 2 weeks and then stored in 0.01 moi/L phos­ © American Society of Neuroradiology phate buffer at 4°C. With a shoulder and cervical surface 1895 18g6 POST AJNR: 15, November 1gg4 coil, plain MR images were then acquired on a 1.5-T unit within 24 hours of symptom onset. All three with a 3-mm section thickness and with T1- and T2- underwent surgical evacuation of their iatro­ weighted images and T2-weighted gradient-echo MR se­ genic ASSH, which as demonstrated on the quences. Pulse parameters were as follows: (a) axial Tl ­ imaging studies and documented at surgery weighted, 750/20/4 (repetition time/echo time/ excitations), goo flip angle; (b) sagittal Tl -weighted, 500/ were extensive in all three extending over 12 20/4, goo flip angle; (c) sagittal and coronal T2-weighted, to 14 spinal levels and associated with spinal 2000/80/2, goo flip angle; (d) axial gradient-echo, 600/ cord and cauda equina compression. Small 13/4, 20o flip angle; (e) sagittal gradient-echo, 500/13/ 4 , amounts of subarachnoid hemorrhage and/or 20° flip angle. Images were displayed on an image matrix epidural blood were also found. of 1g2 or 256 X 256. CT scans were then obtained on The postmortem cases were of three infants, these sa m e specimens in the sagittal view with 1.5-mm­ all of whom died directly as the result of severe thick sections. Subsequently, the frozen vertebral col­ trauma. In two of these infants (both 8 months umns with spinal cords in situ were prepared for sec­ of age), the trauma resulted in severe injuries to tioning on a Reichart-Jung Cryomacrocut (Leica, the spine. The cervical spine was the site of Heidelberg , Germany). They were oriented to be sec­ injury in one and the thoracolumbar spine and tioned sagittally and photographed at approximately coccyx the affected site in the other. The mech­ 1-mm intervals. Cryomicrotome findings were then cor­ related with the postmortem MR and CT results to deter­ anism of injury was a motor vehicle accident in mine the precise location and the imaging characteristics the first and child abuse with severe injury to the of the extravascular blood in the spinal canal and the re­ back, abdomen, and head in the second. After lationship of the hematomas to the spinal cord and death, MR, CT, and cryomicrotome analysis in cauda equina. these two infants revealed sizeable acute spinal In three hospitalized patients, plain CT scans with 3.0- subdural hematomas extending over four to six to 5 -mm -thick sections were done. Plain MR scans were spinal levels. In addition, spinal cord injury then acquired within 6 to 36 hours of the CT. They were all (consisting of cord severance in one and conus done on a 1.5-T unit with a surface coil and 4. 7- to 7 -mm­ maceration in the other), vertebral body frac­ thick sections. Pulse parameters were as follows: (a) axial tures (C-3 and C-7 in one and T-12 and the Tl -weighted, 800/16/1, goo flip angle; (b) sagittal Tl­ weighted, 600/20/4, goo flip angle; (c) sagittal proton­ coccyx in the other), disk injuries, and small density and T2-weighted, 2382/ 20,80/2, goo flip angle; collections of epidural and subarachnoid blood (d) axial gradient-echo, 750-1000/18/1, 20° flip angle; were also present. In the third infant (age 5 ( e) sagittal gradient-echo, 600/18/4, 20o flip angle. Field months), also a victim of child abuse, no cord of view was 23 em for axi<!l and 30 em for sagittal images. injury was present. Findings in this child with an Images were displayed on a matrix of 200 to 220 X 256. injury caused by shaking included mild com­ These MR and CT scans were then reviewed and correlated pression fractures ofT-5 and T-8, disk injury (at with the surgical findings to ascertain in what spinal com­ T-7 to T-8), and a one-level small acute spinal partments the hematoma's were residing, how extensive posterior epidural hematoma (at T -6 to T -7), they were, how much neural compression they were asso­ which was not associated with any cord ciated with, and what signal characteristics they had on MR compression. and what density they had on CT. Our imaging results in both the antemortem and postmortem cases of ASSH revealed the following findings. On plain CT the ASSH ap­ Results peared as a hyperdense lesion within the dural The three hospitalized patients were two sac distinct from the adjacent low-density epi­ adults (ages 21 and 35 years) and one child dural fat and osseous structures (Fig 1). Clump­ (age 11 ), the latter leukemic. All three had ab­ ing, loculation, and streaking of blood within the normal clotting studies. In one of the adults, a dural sac were noted. Similarly, on MR, clump­ gunshot-wound victim with a history of alcohol ing, loculation, and occasionally streaking of abuse and a long complicated hospital course, blood were also noted within the dural sac (Fig the coagulopathy was related to liver dysfunc­ 2) distinct from the adjacent bone and epidural tion. In the other adult, who had an encepha­ fat, the latter hyperintense on T1-weighted im­ lopathy, the patient was on anticoagulant ther­ ages and hypointense on gradient-echo images. apy. Acute paraplegia developed in all three An inhomogeneous and variable signal intensity patients after multiple unsuccessful attempts to the ASSH (depending on its exact age) was at lumbar puncture (n = 2) or epidural anes­ seen on all pulse sequences (Figs 3-5). On thesia (n = 1). Imaging studies were initiated T1-weighted images the acute subdural blood AJNR: 15, November 1994 SUBDU RAL HEMATOMA 1897 Fig 1. ASSHs: plain CT appearance. Plain CT scans in two patients with surgi­ cally proved ASSH show hyperdense lesions within the thecal sac (arrows) , loculated predom inantly to the ventral aspect of the spinal canal at L-3 in one patient (A) and to the dorsa l aspect of the spinal canal at L- 1 in the other (B).