Medical Imaging with 3D Computer for early breast cancer diagnosis

T. Hopp, N.V. Ruiter et al.

Karlsruhe Institute of Technology, Institute for Data Processing and Electronic, Karlsruhe, Germany

KIT – University of the State of Baden-Wuerttemberg and National Research Center of the Helmholtz Association www.kit.edu 3D Ultrasound Computer Tomography

WHO 2012: 1,600,000 new breast cancer cases worldwide ~25% of all cancer cases in women

3D Ultrasound Computer Tomography (USCT) for early breast cancer diagnosis … as harmless as diagnostic ultrasound as economical as X-ray as sensitive as MRI

Patient positioning in USCT KIT 3D USCT

2 09.06.2015 T. Hopp | with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics 3D USCT imaging principle

Receiver

Sender p(t)

Pressure

Transmission ReflectionTime t

>10 Mio. signals Image reconstruction Sound speed 40 GB raw data per breast Attenuation Measurement time 10 s to 4 min. Reflection

3 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics

Image reconstruction

Reflection Sound speed Attenuation Phantom image results (CIRS triple modality phantom)

Parallel and multi GPU computing applied Sound speed + attenuation: ~15 min. per volume Reflection: ~2h per volume

Simplified from Greenleaf et al. (1981)

4 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Image fusion

Reflection + Sound speed (qualitative) (quantitative)

Grayscale Color-coded

Reflection + Sound speed + Attenuation (qualitative) (quantitative) (quantitative)

Grayscale Threshold Threshold

Color-coded

5 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Pilot in-vivo study

Aim: Test KIT 3D USCT with 10 patients 1. Test data acquisition and image reconstruction 2. Test display/combination of multimodal images 3. Compare tissue structures with established imaging method

Study performed on 3 days at University Hospital of Jena, Germany MRI images as ground truth

6 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Patient 1: breast implant

Coronal plane Transversal plane

Registered MRI T1-weighted

USCT Reflection

7 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Patient 2: connective tissue structure

Coronal plane Transversal plane

(detail view) Registered MRI T1-weighted

USCT Reflection

8 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Patient 3: inflammatory carcinoma

Coronal plane Sagittal plane Transversal plane

Registered MRI T1 contrast enhancement

USCT Image fusion: Reflection + sound speed (color-coded)

Sound speed

1300 m/s 1600 m/s

9 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Patient 4: multicenter carcinoma

Coronal plane Sagittal plane Transversal plane

Registered MRI T1 contrast enhancement

USCT Image fusion: Reflection + sound speed (color-coded)

threshold: 1500m/s

Sound speed

1300 m/s 1600 m/s

10 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Image segmentation

Artifacts due to sparse aperture Water background Breast  background segmentation essential Breast boundary (skin) (Semi-)automated approach Inner breast structures

3D USCT reflection image with inhomogeneous water background

Slicewise breast outline detection 3D surface fit Segmented 3D volume

Filtering Edge detection Ellipse fit

Segmentation method to separate breast from water background

11 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Image registration

Comparison of USCT and MRI is challenging due to buoyancy  Image registration to estimate spatial correspondence

MRI: slice 97 USCT: slice 150

Biomechanical model Buoyancy simulation

Registered MRI: slice 150

12 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Quantitative imaging evaluation

Application of a classifier to distinguish between tissue types Example based on registration to mammography (Karmanos Cancer Institute)

Mammogram

Segmented Mammogram Ground truth: segmented mammogram

1600 m/s

Sound speed Registered USCT 1300 m/s sound speed image Classification by Support Vector Machine Feature extraction after registration (9 patients, sound speed and attenuation averaged)

13 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Summary & conclusion

First in-vivo images with 3D USCT: it really works! Successful test of data acquisition and image reconstruction with 10 patients

Image segmentation, registration, fusion involved for clinical evaluation First indications: combination of modalities is promising! Pattern recognition for evaluation and computer aided diagnosis

KIT 3D USCT is ready for a larger clinical study (200 patients) Starting in spring 2015 at University Hospital Mannheim

14 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics Thank you!

We acknowledge support of this project by Deutsche Forschungsgemeinschaft (DFG)

Algorithms / Imaging / Image Processing N. V. Ruiter, M. Zapf, R. Dapp, T. Hopp, W.Y. Tan, B. Qin, H. Gemmeke, et al.

Hardware acceleration E. Kretzek, M. Balzer, et al.

Transducers M. Zapf, H. Gemmeke, et al.

DAQ and Hardware D. Tscherniakhovski, A. Menshikov, et al.

Design and Mechanics L. Berger, B. Osswald, T. Piller, W. Frank, et al.

IPE USCT Group Contact: [email protected]

15 09.06.2015 T. Hopp | Medical Imaging with 3D Ultrasound Computer Tomography Institute for Data Processing and Electronics