Sonia Skamene
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Sonia Skamene has a total of 38 co-authors. Here are the top co-authors by number of shared publications:
- Hossein Naseri
Medical Physics Unit McGill University Health Centre Montreal, QC Canada
2 shared publications - John Kildea
Medical Physics Unit, McGill University, DS1.7141, Cedars Cancer Centre, 1001 boul Décarie, Montreal, Quebec, H4A 3J1, CANADA
2 shared publications - Julia Khriguian
Division of Radiation Oncology McGill University Health Centre Montreal, QC Canada
2 shared publications - Mame Daro Faye
Division of Radiation Oncology, McGill University Health Centre, Montreal, QC H4A 3J1, Canada
2 shared publications - Marc David
Division of Radiation Oncology McGill University Health Centre Montreal, QC Canada
2 shared publications - Marwan Tolba
Department of Radiation Oncology, Dalhousie University, QEII Cancer Centre, Nova Scotia Health Authority, Halifax, Canada
2 shared publications
Latest Publications
- 30 April 2026
- Published by Elsevier in Clinical Oncology
- Vol. 55, 104181
Proton Beam Therapy for Cancer in Children and Adolescents: A Retrospective Canadian Paediatric Tumour Consortium Study
- 30 April 2026
- Published by Elsevier in Clinical Oncology
- Vol. 55, 104181
Proton Beam Therapy for Cancer in Children and Adolescents: A Retrospective Canadian Paediatric Tumour Consortium Study
- Open Access
- 22 May 2023
- Published by JMIR Publications Inc. in JMIR AI
- Vol. 2, e44779
A Scalable Radiomics- and Natural Language Processing–Based Machine Learning Pipeline to Distinguish Between Painful and Painless Thoracic Spinal Bone Metastases: Retrospective Algorithm Development and Validation Study
Background: The identification of objective pain biomarkers can contribute to an improved understanding of pain, as well as its prognosis and better management. Hence, it has the potential to improve the quality of life of patients with cancer. Artificial intelligence can aid in the extraction of objective pain biomarkers for patients with cancer with bone metastases (BMs). Objective: This study aimed to develop and evaluate a scalable natural language processing (NLP)– and radiomics-based machine learning pipeline to differentiate between painless and painful BM lesions in simulation computed tomography (CT) images using imaging features (biomarkers) extracted from lesion center point–based regions of interest (ROIs). Methods: Patients treated at our comprehensive cancer center who received palliative radiotherapy for thoracic spine BM between January 2016 and September 2019 were included in this retrospective study. Physician-reported pain scores were extracted automatically from radiation oncology consultation notes using an NLP pipeline. BM center points were manually pinpointed on CT images by radiation oncologists. Nested ROIs with various diameters were automatically delineated around these expert-identified BM center points, and radiomics features were extracted from each ROI. Synthetic Minority Oversampling Technique resampling, the Least Absolute Shrinkage And Selection Operator feature selection method, and various machine learning classifiers were evaluated using precision, recall, F1-score, and area under the receiver operating characteristic curve. Results: Radiation therapy consultation notes and simulation CT images of 176 patients (mean age 66, SD 14 years; 95 males) with thoracic spine BM were included in this study. After BM center point identification, 107 radiomics features were extracted from each spherical ROI using pyradiomics. Data were divided into 70% and 30% training and hold-out test sets, respectively. In the test set, the accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve of our best performing model (neural network classifier on an ensemble ROI) were 0.82 (132/163), 0.59 (16/27), 0.85 (116/136), and 0.83, respectively. Conclusions: Our NLP- and radiomics-based machine learning pipeline was successful in differentiating between painful and painless BM lesions. It is intrinsically scalable by using NLP to extract pain scores from clinical notes and by requiring only center points to identify BM lesions in CT images.
- 22 May 2023
- Published by JMIR Publications Inc. in JMIR AI
- Vol. 2, e44779
Open AccessA Scalable Radiomics- and Natural Language Processing–Based Machine Learning Pipeline to Distinguish Between Painful and Painless Thoracic Spinal Bone Metastases: Retrospective Algorithm Development and Validation Study
Background: The identification of objective pain biomarkers can contribute to an improved understanding of pain, as well as its prognosis and better management. Hence, it has the potential to improve the quality of life of patients with cancer. Artificial intelligence can aid in the extraction of objective pain biomarkers for patients with cancer with bone metastases (BMs). Objective: This study aimed to develop and evaluate a scalable natural language processing (NLP)– and radiomics-based machine learning pipeline to differentiate between painless and painful BM lesions in simulation computed tomography (CT) images using imaging features (biomarkers) extracted from lesion center point–based regions of interest (ROIs). Methods: Patients treated at our comprehensive cancer center who received palliative radiotherapy for thoracic spine BM between January 2016 and September 2019 were included in this retrospective study. Physician-reported pain scores were extracted automatically from radiation oncology consultation notes using an NLP pipeline. BM center points were manually pinpointed on CT images by radiation oncologists. Nested ROIs with various diameters were automatically delineated around these expert-identified BM center points, and radiomics features were extracted from each ROI. Synthetic Minority Oversampling Technique resampling, the Least Absolute Shrinkage And Selection Operator feature selection method, and various machine learning classifiers were evaluated using precision, recall, F1-score, and area under the receiver operating characteristic curve. Results: Radiation therapy consultation notes and simulation CT images of 176 patients (mean age 66, SD 14 years; 95 males) with thoracic spine BM were included in this study. After BM center point identification, 107 radiomics features were extracted from each spherical ROI using pyradiomics. Data were divided into 70% and 30% training and hold-out test sets, respectively. In the test set, the accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve of our best performing model (neural network classifier on an ensemble ROI) were 0.82 (132/163), 0.59 (16/27), 0.85 (116/136), and 0.83, respectively. Conclusions: Our NLP- and radiomics-based machine learning pipeline was successful in differentiating between painful and painless BM lesions. It is intrinsically scalable by using NLP to extract pain scores from clinical notes and by requiring only center points to identify BM lesions in CT images.
- Open Access
- 2 December 2022
- Published by JMIR Publications Inc. in JMIR Preprints
A Scalable Radiomics- and Natural Language Processing–Based Machine Learning Pipeline to Distinguish Between Painful and Painless Thoracic Spinal Bone Metastases: Retrospective Algorithm Development and Validation Study (Preprint)
BACKGROUND The identification of objective pain biomarkers can contribute to an improved understanding of pain, as well as its prognosis and better management. Hence, it has the potential to improve the quality of life of patients with cancer. Artificial intelligence can aid in the extraction of objective pain biomarkers for patients with cancer with bone metastases (BMs). OBJECTIVE This study aimed to develop and evaluate a scalable natural language processing (NLP)– and radiomics-based machine learning pipeline to differentiate between painless and painful BM lesions in simulation computed tomography (CT) images using imaging features (biomarkers) extracted from lesion center point–based regions of interest (ROIs). METHODS Patients treated at our comprehensive cancer center who received palliative radiotherapy for thoracic spine BM between January 2016 and September 2019 were included in this retrospective study. Physician-reported pain scores were extracted automatically from radiation oncology consultation notes using an NLP pipeline. BM center points were manually pinpointed on CT images by radiation oncologists. Nested ROIs with various diameters were automatically delineated around these expert-identified BM center points, and radiomics features were extracted from each ROI. Synthetic Minority Oversampling Technique resampling, the Least Absolute Shrinkage And Selection Operator feature selection method, and various machine learning classifiers were evaluated using precision, recall, F1-score, and area under the receiver operating characteristic curve. RESULTS Radiation therapy consultation notes and simulation CT images of 176 patients (mean age 66, SD 14 years; 95 males) with thoracic spine BM were included in this study. After BM center point identification, 107 radiomics features were extracted from each spherical ROI using pyradiomics. Data were divided into 70% and 30% training and hold-out test sets, respectively. In the test set, the accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve of our best performing model (neural network classifier on an ensemble ROI) were 0.82 (132/163), 0.59 (16/27), 0.85 (116/136), and 0.83, respectively. CONCLUSIONS Our NLP- and radiomics-based machine learning pipeline was successful in differentiating between painful and painless BM lesions. It is intrinsically scalable by using NLP to extract pain scores from clinical notes and by requiring only center points to identify BM lesions in CT images.
- 2 December 2022
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessA Scalable Radiomics- and Natural Language Processing–Based Machine Learning Pipeline to Distinguish Between Painful and Painless Thoracic Spinal Bone Metastases: Retrospective Algorithm Development and Validation Study (Preprint)
BACKGROUND The identification of objective pain biomarkers can contribute to an improved understanding of pain, as well as its prognosis and better management. Hence, it has the potential to improve the quality of life of patients with cancer. Artificial intelligence can aid in the extraction of objective pain biomarkers for patients with cancer with bone metastases (BMs). OBJECTIVE This study aimed to develop and evaluate a scalable natural language processing (NLP)– and radiomics-based machine learning pipeline to differentiate between painless and painful BM lesions in simulation computed tomography (CT) images using imaging features (biomarkers) extracted from lesion center point–based regions of interest (ROIs). METHODS Patients treated at our comprehensive cancer center who received palliative radiotherapy for thoracic spine BM between January 2016 and September 2019 were included in this retrospective study. Physician-reported pain scores were extracted automatically from radiation oncology consultation notes using an NLP pipeline. BM center points were manually pinpointed on CT images by radiation oncologists. Nested ROIs with various diameters were automatically delineated around these expert-identified BM center points, and radiomics features were extracted from each ROI. Synthetic Minority Oversampling Technique resampling, the Least Absolute Shrinkage And Selection Operator feature selection method, and various machine learning classifiers were evaluated using precision, recall, F1-score, and area under the receiver operating characteristic curve. RESULTS Radiation therapy consultation notes and simulation CT images of 176 patients (mean age 66, SD 14 years; 95 males) with thoracic spine BM were included in this study. After BM center point identification, 107 radiomics features were extracted from each spherical ROI using pyradiomics. Data were divided into 70% and 30% training and hold-out test sets, respectively. In the test set, the accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve of our best performing model (neural network classifier on an ensemble ROI) were 0.82 (132/163), 0.59 (16/27), 0.85 (116/136), and 0.83, respectively. CONCLUSIONS Our NLP- and radiomics-based machine learning pipeline was successful in differentiating between painful and painless BM lesions. It is intrinsically scalable by using NLP to extract pain scores from clinical notes and by requiring only center points to identify BM lesions in CT images.
- Open Access
- 15 February 2022
- Published by American Association for Cancer Research (AACR) in Cancer Research
- Vol. 82 (4_Suppleme) , PD7-07
Abstract PD7-07: Surgical management and contralateral breast cancer risk in women with a history of radiation therapy for Hodgkin lymphoma: Results from a population-based cohort
Background: Women with a history of chest radiation for Hodgkin lymphoma are at an increased risk of developing breast cancer. Although mastectomy has historically been recommended for surgical treatment of breast cancers in women with prior radiation exposure, surgical management trends and contralateral breast cancer risk in this population remain undefined. Methods: We performed a population-based retrospective study using the Surveillance, Epidemiology, and End Results (SEER) database from 1990-2016. Our cohort included women who received radiation for Hodgkin lymphoma prior to 30 years old and were diagnosed with a subsequent breast cancer. We evaluated trends in local therapy including rates of breast conserving surgery (BCS) and mastectomy. In those undergoing unilateral surgery, the Kaplan-Meier method was used to estimate the 5- and 10-year cumulative incidence of contralateral breast cancer. Results: Our final cohort included 295 women with a median age of 22 years (range, 8-30 years) at Hodgkin lymphoma diagnosis, and 42 years (range, 22-65 years) at breast cancer diagnosis. Of these patients, 263 (89.2%) presented with unilateral breast cancer, while 32 (10.8%) presented with synchronous bilateral breast cancer. Overall, BCS was performed in 17.3% of patients and mastectomy was performed in 82.7%. In the 263 patients presenting with unilateral breast cancer, 50 (19.0%) underwent BCS and 213 (81.0%) underwent mastectomy. Subgroup analysis of mastectomy patients with surgical laterality information available demonstrated a 40.5% bilateral mastectomy rate. In the entire cohort, the 5-year incidence of contralateral breast cancer in women who underwent unilateral surgery was 9.4% (95% CI, 5.6-15.4), increasing to 20.2% (95% CI, 13.7-29.2) at 10-years of follow up. Hormone receptor status of the index breast cancer was not associated with significant differences in the incidence of contralateral breast cancer (p=0.13). Conclusions: Women with a history of prior chest radiation for Hodgkin Lymphoma with a diagnosis with breast cancer have a 10-year contralateral breast cancer risk of 20%. These findings support consideration of contralateral prophylactic mastectomy during surgical decision-making in this high-risk patient population. Citation Format: Elisheva R Eisenberg, Anna Weiss, Ipshita Prakash, Sonia Skamene, Mark Basik, Jean Francois Boileau, Lissa Ajjamada, Michael Pollak, Stephanie M Wong. Surgical management and contralateral breast cancer risk in women with a history of radiation therapy for Hodgkin lymphoma: Results from a population-based cohort [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr PD7-07.
- 15 February 2022
- Published by American Association for Cancer Research (AACR) in Cancer Research
- Vol. 82 (4_Suppleme) , PD7-07
Open AccessAbstract PD7-07: Surgical management and contralateral breast cancer risk in women with a history of radiation therapy for Hodgkin lymphoma: Results from a population-based cohort
Background: Women with a history of chest radiation for Hodgkin lymphoma are at an increased risk of developing breast cancer. Although mastectomy has historically been recommended for surgical treatment of breast cancers in women with prior radiation exposure, surgical management trends and contralateral breast cancer risk in this population remain undefined. Methods: We performed a population-based retrospective study using the Surveillance, Epidemiology, and End Results (SEER) database from 1990-2016. Our cohort included women who received radiation for Hodgkin lymphoma prior to 30 years old and were diagnosed with a subsequent breast cancer. We evaluated trends in local therapy including rates of breast conserving surgery (BCS) and mastectomy. In those undergoing unilateral surgery, the Kaplan-Meier method was used to estimate the 5- and 10-year cumulative incidence of contralateral breast cancer. Results: Our final cohort included 295 women with a median age of 22 years (range, 8-30 years) at Hodgkin lymphoma diagnosis, and 42 years (range, 22-65 years) at breast cancer diagnosis. Of these patients, 263 (89.2%) presented with unilateral breast cancer, while 32 (10.8%) presented with synchronous bilateral breast cancer. Overall, BCS was performed in 17.3% of patients and mastectomy was performed in 82.7%. In the 263 patients presenting with unilateral breast cancer, 50 (19.0%) underwent BCS and 213 (81.0%) underwent mastectomy. Subgroup analysis of mastectomy patients with surgical laterality information available demonstrated a 40.5% bilateral mastectomy rate. In the entire cohort, the 5-year incidence of contralateral breast cancer in women who underwent unilateral surgery was 9.4% (95% CI, 5.6-15.4), increasing to 20.2% (95% CI, 13.7-29.2) at 10-years of follow up. Hormone receptor status of the index breast cancer was not associated with significant differences in the incidence of contralateral breast cancer (p=0.13). Conclusions: Women with a history of prior chest radiation for Hodgkin Lymphoma with a diagnosis with breast cancer have a 10-year contralateral breast cancer risk of 20%. These findings support consideration of contralateral prophylactic mastectomy during surgical decision-making in this high-risk patient population. Citation Format: Elisheva R Eisenberg, Anna Weiss, Ipshita Prakash, Sonia Skamene, Mark Basik, Jean Francois Boileau, Lissa Ajjamada, Michael Pollak, Stephanie M Wong. Surgical management and contralateral breast cancer risk in women with a history of radiation therapy for Hodgkin lymphoma: Results from a population-based cohort [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr PD7-07.