top of page

Clinical Evaluation

A thoughtful combination of advanced model-based, machine learning, and deep learning methods coupled with advanced registration between the baseline and follow-up scans yields results that are superior to standalone analysis.  The method presents trade-offs between the type of expert knowledge available, the amount and representativeness of data sets, and the availability of tagged data sets. iCAS follows a unique bootstrapping approach, which consists of generating radiologist-validated tagged datasets that are then used by AI and Deep Learning methods to improve their precision and increase their specificity and coverage.

TM

iStock-1074166394Post.jpg

The proposed research is part of a longstanding ongoing multidisciplinary collaboration.  The team has developed methods for the follow-up of liver tumors [1-3], lung tumors [4], liver metastases [5-6], and brain tumors [7-10] in CT and MRI scans. For example, our results on personalized follow-up liver tumor burden quantification using model-based and convolutional neural networks (CNN) methods in longitudinal CT scans of 222 tumors from 31 patients yield an average volume overlap error of 17% (std=11.2) far better than the 28.5% (std=22.5) of stand-alone segmentation. Importantly, the robustness of our method improved from 67% for stand-alone global CNN segmentation to 100% [1]. Unlike other medical imaging deep learning approaches, which require large annotated training datasets, our methods exploit the follow-up framework to yield accurate tumor tracking and failure detection and correction with a small training dataset. We have obtained similar results for brain, kidney and lung tumors.  A prototype software for the follow-up of optic path gliomas [9] is currently in use at a medical center in Israel.

  1. Patient-specific Convolutional Neural Networks for robust automatic liver tumor delineation in longitudinal CT studies.  R. Vivanti, L. Joskowicz, A. Ephrat, N. Lev-Cohain, J. Sosna. Medical and Biological Engineering and Computing, Springer. Published online, March 2018.

  2. Computer-based radiological longitudinal evaluation of meningiomas following stereotactic radiosurgery. E. Ben Shimol, L. Joskowicz, R. Eliahou, Y. Shoshan. Int. J. of Computer Aided Surgery and Radiology 13(2):215-228, 2018.

  3. A fully automatic end-to-end method for content-based image retrieval of CT scans with similar liver lesion annotations. A. Spanier, N. Caplan, J. Sosna, B. Acar, L. Joskowicz. Int. J. of Computer Aided Surgery and Radiology 13(1):165–174, 2018.

  4. Automatic detection of new tumors and tumor burden evaluation in longitudinal liver CT scan studies. R. Vivanti, A. Szeskin, L. Joskowicz, N. Lev-Cohain, J. Sosna. Int. J. of Computer-Aided Radiology and Surgery 12(11):1945-1957, 2017.

  5. Automatic lung tumor segmentation with leaks removal in follow-up CT studies. R. Vivanti, O. Karaaslan, L. Joskowicz, J. Sosna. Int. J. Computer Aided Radiology and Surgery 10(9):1505-14, 2015.

  6. Quantitative functional MRI biomarkers improved early detection of colorectal liver metastases. Y. Edrei, M. Freiman, M. Sklair, G. Tsarfati, E. Gross, L. Joskowicz, R. Abramovitch. J. Magnetic Resonance Imaging 39(5):1246-53, 2014.

  7. The effect of chemotherapy on optic pathway gliomas and their sub components: a volumetric MR analysis study. B. Shofty, L. Weizman, L. Joskowicz, L. Pratt, R. Dvir, L. Ravid, D. Ben Bashat, M. Ilon, A. Kessler, Pediatric Blood & Cancer 62(8):1353-9, 2015.

  8. Tumor burden evaluation in NF1 patients with plexiform neurofibromas in the daily clinical practice. L. Pratt, D. Helfer, L. Weizman, B. Shofty, S. Constantini, L. Joskowicz, D. Ben Bashat, L. Ben-Sira. Acta Neurochirurgica 157(5):855--861, 2015.

  9. Segmentation and follow-up of multi-component low-grade gliomas in longitudinal MRI studies. L. Weizman, D. Ben Bashat, L. Joskowicz, D. Rubin, K.W. Yeom, S. Constantini, B. Shofty L. Ben-Sira. Medical Physics 41:052303, 2014.

  10. Segmentation and follow-up of multi-component low-grade gliomas in longitudinal MRI studies.  L. Weizman, D. Ben Bashat, L. Joskowicz, D. Rubin, K.W. Yeom, S. Constantini, B. Shofty L. Ben-Sira. Medical Physics 41:052303, 2014.

bottom of page