Doctor of Philosophy
Ward, Aaron D.
Multi-parametric magnetic resonance imaging (mp-MRI) is emerging as a useful tool for classifying prostate cancer (PCa); however, it suffers from two major limitations: (1) complex, multi-dimensional signals make interpretation challenging and (2) inter-observer variability of lesion classification between physicians. Critically needed are methods for augmenting the interpretability of mp-MRI to assist in lesion classification. To meet this need, we leveraged a patient cohort with post-surgery pathologist-annotated transverse histology images registered to pre-surgery in-vivo mp-MRI with a measured target registration error. We developed a radiomics-based machine learning model trained on annotations for PCa vs. non-PCa, and found that a 5-feature Naïve-Bayes classifier classified these two types of regions in leave-one-patient-out cross-validation with an area under the receiver operating characteristic curve of 0.80. We then investigated augmentation of mp-MRI interpretability via additional imaging using prostate-specific membrane antigen positron emission tomography (PSMA-PET). Using patients imaged on a PET/MRI hybrid-scanner, we investigated segmentation recommendations for classifying the dominant intraprostatic lesion. For focal therapy, we found an optimal threshold of 67% SUVmax and an 8.4 mm margin produced mean voxel-wise sensitivity of 95% with mean specificity 76%. For guided biopsy, we found that a threshold of 81% SUVmax and a 5.2 mm margin produced a mean sensitivity of 65% and mean specificity of 95%. Furthermore, we assessed the effect of boosting high-dose-rate brachytherapy treatment plans towards these segmentations. We found that dose to the underlying high-grade cancer was significantly increased in targeted plans, compared to whole-gland plans, while maintaining all dose constraints.
Summary for Lay Audience
Prostate cancer (PCa) is the most common non-skin cancer in men in Canada. PCa is commonly diagnosed by extracting prostate tissue through biopsy needles which are guided with ultrasound, known as trans-rectal ultrasound (TRUS)-guided biopsy. However, TRUS-guided biopsy isn’t very accurate, and many cancers may be missed. Multi-parametric magnetic resonance imaging (mp-MRI) is another form of imaging that has shown to improve detection of PCa. However, mp-MRI has two major challenges: (1) it is complex and thus even expert physicians have difficulty analyzing the images, and (2) there is a lot of variability between interpretations made by different physicians for the same image. Critically needed are methods assisting physicians on interpreting mp-MRI to assist in classifying PCa. To meet these needs, we use a dataset of men imaged with mp-MRI prior to surgical prostate removal. These prostates were sent to pathology where they were sliced, and cancer was annotated by the pathologists. These histology slices were mapped back to the pre-surgery mp-MRI so we could determine the true location of the underlying disease. We developed a machine learning model that was able to accurately distinguish PCa from non-PCa based on the appearance of the cancer on mp-MRI. Furthermore, we investigated another method of assisting physicians with mp-MRI interpretation using additional imaging known prostate-specific membrane antigen positron emission tomography (PSMA-PET). Using a subset of 12 patients imaged on a PET/MRI scanner that acquired mp-MRI and PET images at the same time, we investigated recommendations for contouring the most threatening lesion in the prostate, known as the dominant intraprostatic lesion. Furthermore, using these contours generated on PSMA-PET, we aimed to determine the effect of focusing dose in brachytherapy treatment plans towards these contours in comparison to what is commonly performed in the clinic, namely whole-gland plans. We found that the highly aggressive and life-threatening cancer received a significantly increased dose in these targeted plans, compared to whole-gland plans, which may provide improved cancer treatment.
Alfano, Ryan M., "Computer-assisted lesion classification and intervention planning for prostate cancer" (2021). Electronic Thesis and Dissertation Repository. 7831.
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.