Our Artificial Intelligence Group aims to move towards personalized healthcare for patients treated with radiotherapy using deep learning methods.
Treatment Outcome Prediction
Cancer is the leading cause of death in Canada. Radiotherapy is used in about half of all cancer treatments. However, it is clinically challenging to predict which patients will benefit from which treatment combination despite improved protocols, imaging techniques for cancer management, and combining various radiotherapy treatment modalities. An accurate method for predicting a patient’s likelihood of response may reduce unnecessary interventions, lower healthcare costs, and reduce side effects. Therefore, investigating how pre-treatment patient characteristics influence treatment efficacy as measured by post-treatment response is crucial. The aim of Yujing and Luca’s work is to develop a patient-specific, machine learning-based multi-modal treatment outcome prediction model for patients suffering from gynecological cancers. Our group has recently shown that the patient-specific radiation dose response may be influenced by inter-patient variation in tumor nuclei size. Therefore, the models developed by Yujing and Luca model will not only integrate diagnostic images such as Computed Tomography (CT), Magnetic Resonance (MR), and Ultrasound (US) images to detect high-order features that determine treatment response on a patient-specific basis, but also tumor nucleus size and cell spacing data obtained from scanned digital images of histopathology slides. Correlation between cell morphology and treatment outcome will also be investigated
Treatment Outcome Prediction
Alana and Luca are investigating the inter-observer variability of the manual segmentation of tumor regions in endoscopy images and its effect on treatment outcomes. Furthermore, they are developing a deep learning-based segmentation tool that can learn from multiple observers labels with high inter-observer variability.
AI-based dosimetry and catheter reconstruction for brachytherapy application
The precalculated dose used by clinical treatment planning systems is based on the American Association of Physicists in Medicine (AAPM) Task Group 43 (TG-43) formalism which describes dose deposition around a single source centrally positioned in a spherical water phantom with unit density. The influence of patient tissue and applicator heterogeneities, intersource attenuation, and finite patient dimensions are ignored. Recently, more advanced model based dose calculation algorithms (MBDCAs) calculating dose to medium in medium have been developed. In MBDCAs, dose calculations are performed using the patient’s CT or MR images. Voxel-by-voxel assignment of tissue mass density and elemental composition is required. Tissue mass density is obtained from CT or synthetic CT images using a Hounsfield Unit (HU) to density calibration curve. Tissue composition and nominal mass density can also be assigned to contoured organs. Exact description of the source and applicator geometry, material composition, and nominal mass density is required as well. MBDCAs such as the Monte Carlo method, provide a detailed and accurate method for calculation of absorbed dose in heterogeneous systems such as the human body however are too time consuming to be used in a clinical workflow.
The main goal of Sébastien’s PhD project is to develop a precise and automated dosimetry algorithm that will take into account patient’s tissue/applicator heterogeneities and replace the time consuming Monte-Carlo simulations. Sébastien is also working on automating patient organ/tumor segmentation as well as catheter/applicator reconstruction to decrease the total treatment time.
Prediction of Radiation Induced Toxicity in High Dose Rate Brachytherapy Treatment of Breast Cancer
The aim of Marie’s current project is to develop a deep learning-based algorithm that can predict recurrence and potential long-term side effects based on a variety of patient-specific parameters for breast cancer patients treated with high dose rate brachytherapy.
Artificial intelligence with radiomics and deep learning for the diagnosis of malignancy in pancreatic cysts: a pilot study
Pancreatic cysts are common, with a prevalence of 15% in the general population, and up to 37% in older populations. While some cystic lesions of the pancreas, such as serous cystadenomas, pseudocysts and epidermoid cysts, have little to no malignant potential, others including intraductal papillary mucinous neoplasm (IPMN) and mucinous cystic neoplasm carry a risk of invasive adenocarcinoma. Even among the potentially malignant cyst types, it is difficult to predict which cysts harbor malignancy or have a high risk of malignant transformation. In this project, we aim to 1) develop a deep learning algorithm using radiomics from EUS, CT and MR images of pancreas cysts to predict malignancy; 2) To assess the diagnostic ability of this prediction algorithm in differentiating malignant from benign pancreatic cysts; 3) To compare the diagnostic ability of this AI algorithm to standard of care criteria for predicting malignant pancreatic cystic neoplasms; 4) To assess feasibility of the current methodology and AI algorithm for a large multicenterd trial with an external validation cohort.
Auto-segmentation of brain structures
Radiation therapy to brain tumor requires a delicate balance between delivering adequate tumorcidal dose to the tumor lesion while minimizing radiation to healthy organs. Segmentation of organs at risk (OARs) is time-consuming and subject to inter-observer variability. Auto-segmentation can 1) increase workflow output by lightening segmentation time, and 2) decrease inter-observer variability. The Radiation Oncology Department at the McGill University Health Centre in collaboration with the Montreal Children Hospital treat pediatric patients with brain tumors. Compared to adult population, radiation therapy to pediatric population is higher stake for several reasons.1) As opposed to adult patients with mature anatomy, radiation can several impair normal growth of pediatric patients; 2) There are more brain structures in pediatric patients to consider; 3) Most children brain structures are more sensitive to radiation than those of adults. Thus, accurate segmentation is of great importance. The project is to investigate deep learning algorithms that can accurately auto-segment OARs in pediatric patients undergoing brain radiation therapy.
from McGill University in 2021.
ESTRO Annual meeting, 27.08.2021.
from MITACS in 2021.
Inter-Observer Variability and Deep Learning in Rectal Tumor Segmentation from Endoscopy Images Presentation
The COMP Annual Scientific Meeting 2021, 22.06.2021.
In: arXiv:2105.01824 [physics], 2021, (arXiv: 2105.01824).
Treatment outcome Prediction for gynecological cancers patients with a machine learning model using pre/post diagnostic image modalities and digital histopathology images Presentation
CRUK RadNet Manchester AI for Optimising Radiotherapy Outcomes Workshop, 10.02.2021.
Fire-Up - Radiation Treatment Outcome Prediction Presentation
Fire-Up Presentation, 09.02.2021.
Approaching automated applicator digitization from a new angle: using sagittal images to improve deep learning accuracy and robustness in high-dose-rate prostate brachytherapy Journal Article
In: 2021 ABS Annual Meeting, 2021, (Type: Journal Article).
In: World Congress of Brachytherapy 2021, 2021, (Type: Journal Article).
In: International Journal of Radiation Oncology, Biology, Physics, 108 (3), pp. 802–812, 2020, ISSN: 1879-355X.
12.06.2020, (Type: Journal Article).
Math And Physics Class Of 1965 Prize Award
from McGill University in 2020.
In: Int J Radiat Oncol Biol Phys, 108 (3), pp. 802-812, 2020, ISSN: 0360-3016.
In: 2020 Joint AAPM textbar COMP Virtual Meeting, 2020, (Type: Journal Article).