Michael Hardisty
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
Research Subjects
Michael Hardisty has a total of 91 co-authors. Here are the top co-authors by number of shared publications:
- Cari Marisa Whyne
Faculty of Engineering, Institute of Biomedical Engineering, University of Toronto, Toronto, Canada
16 shared publications - David Burns
Orthopaedic Biomechanics Lab, Holland Bone and Joint Program, Sunnybrook Research Institute, Toronto, ON M4N 3M5, Canada
7 shared publications Cari M. Whyne
Orthopaedic Biomechanics Laboratory, Sunnybrook Research Institute, 2075 Bayview Avenue, Toronto, ON, M4N 3M5, Canada; Institute of Biomedical Engineering, Faculty of Engineering, University of Toronto, 164 College Street, Toronto, ON, M5S 3G9, Canada; Division of Orthopaedic Surgery, Department of Surgery, University of Toronto, 149 College Street, Toronto, ON, M5T 1P5, Canada
4 shared publicationsCari Whyne
4 shared publications- Helen Razmjou
Bone and Joint Program, Holland Orthopedic & Arthritic Centre, Sunnybrook Health Sciences Centre
4 shared publications - Stewart McLachlin
Department of Mechanical and Mechatronics Engineering, University of Waterloo, Waterloo, ON N2L 3G1, Canada
4 shared publications
Latest Publications
- Open Access
- 30 October 2026
- Published by MDPI AG in Journal of Clinical Medicine
- Vol. 15 (21) , 8453
CT-Defined Osteosarcopenia Is Associated with Mortality in Metastatic Prostate Cancer: A Longitudinal Analysis of Opportunistic CT Imaging
Background/Objectives: In metastatic prostate cancer, loss of bone density (osteoporosis), muscle quantity and quality (sarcopenia) may contribute to fractures, falls, and mortality. Unfortunately, their longitudinal assessment is poorly characterized. In prostate cancer, ageing, advanced disease, and cancer treatments may accelerate bone and muscle loss. This study assessed longitudinal computed tomography-derived musculoskeletal biomarkers in metastatic prostate cancer and the association between baseline osteosarcopenia and mortality. Methods: This retrospective single-centre cohort included 437 patients, 3595 muscle computed tomography (CT) assessments and 2944 bone CT assessments. An automated deep-learning pipeline quantified lumbar trabecular bone mineral density (BMD), psoas muscle volume (PV), and psoas muscle density. Osteosarcopenia was defined as concurrent osteoporosis (BMD < 135 Hounsfield Units (HU)) and sarcopenia (volumetric psoas muscle index (vPMI) < 30). Patient-aware mixed-effects models characterized biomarker trends and reverse-time analysis. Kaplan–Meier analysis and Cox proportional-hazards regressions were utilized to assess mortality. Results: Baseline osteosarcopenia affected 75 patients (17.2%). Psoas density decreased by 1.569 HU/year (95% confidence interval (CI) −1.703 to −1.434 HU/yr). Nonlinear models estimated mean declines in psoas volume (−6.42 cm3/year) and bone density (−6.01 HU/year). Among deceased patients, modelled final-year declines showed −12.57% in psoas density and −10.28% in psoas volume. Baseline osteosarcopenia was associated with higher all-cause mortality (Hazard Ratio (HR) 1.611; 95% CI 1.134–2.288; p = 0.008). Conclusions: Osteosarcopenia is common and progressive in metastatic prostate cancer, and muscle loss accelerates near death. Baseline osteosarcopenia independently identifies patients at increased mortality risk. CT imaging provides a scalable approach for longitudinal assessment of musculoskeletal health and offers clinically important information without additional patient burden.
- 30 October 2026
- Published by MDPI AG in Journal of Clinical Medicine
- Vol. 15 (21) , 8453
Open AccessCT-Defined Osteosarcopenia Is Associated with Mortality in Metastatic Prostate Cancer: A Longitudinal Analysis of Opportunistic CT Imaging
Background/Objectives: In metastatic prostate cancer, loss of bone density (osteoporosis), muscle quantity and quality (sarcopenia) may contribute to fractures, falls, and mortality. Unfortunately, their longitudinal assessment is poorly characterized. In prostate cancer, ageing, advanced disease, and cancer treatments may accelerate bone and muscle loss. This study assessed longitudinal computed tomography-derived musculoskeletal biomarkers in metastatic prostate cancer and the association between baseline osteosarcopenia and mortality. Methods: This retrospective single-centre cohort included 437 patients, 3595 muscle computed tomography (CT) assessments and 2944 bone CT assessments. An automated deep-learning pipeline quantified lumbar trabecular bone mineral density (BMD), psoas muscle volume (PV), and psoas muscle density. Osteosarcopenia was defined as concurrent osteoporosis (BMD < 135 Hounsfield Units (HU)) and sarcopenia (volumetric psoas muscle index (vPMI) < 30). Patient-aware mixed-effects models characterized biomarker trends and reverse-time analysis. Kaplan–Meier analysis and Cox proportional-hazards regressions were utilized to assess mortality. Results: Baseline osteosarcopenia affected 75 patients (17.2%). Psoas density decreased by 1.569 HU/year (95% confidence interval (CI) −1.703 to −1.434 HU/yr). Nonlinear models estimated mean declines in psoas volume (−6.42 cm3/year) and bone density (−6.01 HU/year). Among deceased patients, modelled final-year declines showed −12.57% in psoas density and −10.28% in psoas volume. Baseline osteosarcopenia was associated with higher all-cause mortality (Hazard Ratio (HR) 1.611; 95% CI 1.134–2.288; p = 0.008). Conclusions: Osteosarcopenia is common and progressive in metastatic prostate cancer, and muscle loss accelerates near death. Baseline osteosarcopenia independently identifies patients at increased mortality risk. CT imaging provides a scalable approach for longitudinal assessment of musculoskeletal health and offers clinically important information without additional patient burden.
- 28 October 2026
- Published by Elsevier in Ultrasound in Medicine & Biology
- Vol. 53 (1) , 237-245
HipSAFE: detecting hip fractures on ultrasound imaging using deep learning in a porcine cadaveric model
- 28 October 2026
- Published by Elsevier in Ultrasound in Medicine & Biology
- Vol. 53 (1) , 237-245
HipSAFE: detecting hip fractures on ultrasound imaging using deep learning in a porcine cadaveric model
- Open Access
- 29 September 2026
- Published by MDPI AG in Preprints.org
Osteosarcopenia Predicts Mortality in Metastatic Prostate Cancer: A Longitudinal Analysis of Opportunistic CT Imaging
Background/Objectives: Osteoporosis and sarcopenia are associated with fractures, falls, and mortality. In prostate cancer, aging, advanced disease, and androgen deprivation therapy may accelerate concurrent bone and muscle loss. This study characterized longitudinal CT-derived musculoskeletal biomarkers in metastatic prostate cancer and evaluated the association between baseline osteosarcopenia and overall survival. Methods: This retrospective single-centre cohort included patients treated from 2009 to 2022, with follow-up imaging through 2024. An automated deep-learning pipeline quantified lumbar trabecular bone mineral density (BMD), psoas muscle volume (PV), psoas muscle density (PD), and volumetric psoas muscle index (vPMI). Osteosarcopenia was defined as concurrent osteoporosis (BMD < 135 HU) and sarcopenia (vPMI < 30). Survival from baseline CT was assessed using Kaplan-Meier analysis and multivariable Cox regression. Results: Among 437 patients with 3389 CT scans, osteoporosis, sarcopenia, and osteosarcopenia increased from 29.6%, 46.9%, and 17.2% at baseline to 54.0%, 82.8%, and 45.5% at final imaging. BMD declined by 2.1-3.7% annually, PD by 3.2-3.9%, and PV by 3.6-4.6% (all p < 0.0001). In the final year of life, PD declined by 12.0% and PV by 9.3%. Baseline osteosarcopenia remained independently associated with mortality after adjustment for age, time from diagnosis to baseline CT, PSA, and Gleason risk group (HR 2.11, 95% CI 1.50-2.99; p < 0.001). Conclusions: Osteosarcopenia is common and progressive in metastatic prostate cancer, and muscle loss accelerates near death. Baseline osteosarcopenia independently identifies patients at increased mortality risk. Automated analysis of routine surveillance CT may enable longitudinal musculoskeletal assessment and prognostic risk stratification.
- 29 September 2026
- Published by MDPI AG in Preprints.org
Open AccessOsteosarcopenia Predicts Mortality in Metastatic Prostate Cancer: A Longitudinal Analysis of Opportunistic CT Imaging
Background/Objectives: Osteoporosis and sarcopenia are associated with fractures, falls, and mortality. In prostate cancer, aging, advanced disease, and androgen deprivation therapy may accelerate concurrent bone and muscle loss. This study characterized longitudinal CT-derived musculoskeletal biomarkers in metastatic prostate cancer and evaluated the association between baseline osteosarcopenia and overall survival. Methods: This retrospective single-centre cohort included patients treated from 2009 to 2022, with follow-up imaging through 2024. An automated deep-learning pipeline quantified lumbar trabecular bone mineral density (BMD), psoas muscle volume (PV), psoas muscle density (PD), and volumetric psoas muscle index (vPMI). Osteosarcopenia was defined as concurrent osteoporosis (BMD < 135 HU) and sarcopenia (vPMI < 30). Survival from baseline CT was assessed using Kaplan-Meier analysis and multivariable Cox regression. Results: Among 437 patients with 3389 CT scans, osteoporosis, sarcopenia, and osteosarcopenia increased from 29.6%, 46.9%, and 17.2% at baseline to 54.0%, 82.8%, and 45.5% at final imaging. BMD declined by 2.1-3.7% annually, PD by 3.2-3.9%, and PV by 3.6-4.6% (all p < 0.0001). In the final year of life, PD declined by 12.0% and PV by 9.3%. Baseline osteosarcopenia remained independently associated with mortality after adjustment for age, time from diagnosis to baseline CT, PSA, and Gleason risk group (HR 2.11, 95% CI 1.50-2.99; p < 0.001). Conclusions: Osteosarcopenia is common and progressive in metastatic prostate cancer, and muscle loss accelerates near death. Baseline osteosarcopenia independently identifies patients at increased mortality risk. Automated analysis of routine surveillance CT may enable longitudinal musculoskeletal assessment and prognostic risk stratification.
- Open Access
HipSAFE: automating hip fracture detection on ultrasound imaging using deep learning
Falls among older adults can result in hip fractures that requires x-ray based assessment at emergency department (ED). Only 25.7% of patients presenting to EDs are diagnosed with a hip fracture, as such improved diagnosis prior to transportation to hospital could result in fewer hospital visits and improved triaging. Patient with hip fracture could be immediately directed to centres with orthopaedic surgeons, allowing for reduced time-to-surgery, particularly in rural communities. Ultrasound (US) imaging is portable and can identify fractures but requires expertise, particularly related to image interpretation. Deep learning may reduce operator dependence by automating image interpretation. This study aims to develop HipSAFE, a hip fracture detection tool on US, to support triaging by nurses and paramedics. We hypothesize that diagnostic accuracy will be comparable to pelvic x-ray diagnostic performance in a preclinical study. Bilateral hind limbs of 15 porcine cadavers were imaged by US-naïve operators before and after an iatrogenic hip fracture. The limbs were divided into training, validation, and test (8 femurs) sets. The training data were augmented (geometric and photometric transformations). The models included MobileNetV3 (S/L), EfficientNet-Lite (0–2), and ResNet (18/50). Using a moving average aggregation on the operator cine clips, EfficientNet-Lite0 achieved the highest performance (F1=0.944 [95% CI:0.880-0.987]; sensitivity=89.5% [78.6-97.5%]; specificity = 100.0% [100.0-100.0]). The majority voting ensemble model ranked second (F1=0.932 [0.857-0.984]). Naïve operators and radiologists had lower performance (F1=0.667 [0.596-0.758] and 0.685 [0.597-0.729]). This pre-clinical study demonstrated that HipSAFE has excellent diagnostic accuracy and there may be a role for US in improving hip trauma triaging, especially for rural and resource-constrained environments.
Open AccessHipSAFE: automating hip fracture detection on ultrasound imaging using deep learning
Falls among older adults can result in hip fractures that requires x-ray based assessment at emergency department (ED). Only 25.7% of patients presenting to EDs are diagnosed with a hip fracture, as such improved diagnosis prior to transportation to hospital could result in fewer hospital visits and improved triaging. Patient with hip fracture could be immediately directed to centres with orthopaedic surgeons, allowing for reduced time-to-surgery, particularly in rural communities. Ultrasound (US) imaging is portable and can identify fractures but requires expertise, particularly related to image interpretation. Deep learning may reduce operator dependence by automating image interpretation. This study aims to develop HipSAFE, a hip fracture detection tool on US, to support triaging by nurses and paramedics. We hypothesize that diagnostic accuracy will be comparable to pelvic x-ray diagnostic performance in a preclinical study. Bilateral hind limbs of 15 porcine cadavers were imaged by US-naïve operators before and after an iatrogenic hip fracture. The limbs were divided into training, validation, and test (8 femurs) sets. The training data were augmented (geometric and photometric transformations). The models included MobileNetV3 (S/L), EfficientNet-Lite (0–2), and ResNet (18/50). Using a moving average aggregation on the operator cine clips, EfficientNet-Lite0 achieved the highest performance (F1=0.944 [95% CI:0.880-0.987]; sensitivity=89.5% [78.6-97.5%]; specificity = 100.0% [100.0-100.0]). The majority voting ensemble model ranked second (F1=0.932 [0.857-0.984]). Naïve operators and radiologists had lower performance (F1=0.667 [0.596-0.758] and 0.685 [0.597-0.729]). This pre-clinical study demonstrated that HipSAFE has excellent diagnostic accuracy and there may be a role for US in improving hip trauma triaging, especially for rural and resource-constrained environments.
- Open Access
- 20 February 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (4) , 2084
Recurrence with Correlation Network for Medical Image Registration
This work presents Recurrence with Correlation Network(RWCNet), a novel multi-scale recurrent neural network architecture for medical image registration that integrates core principles from optical flow, including correlation volume computation and inference-time instance optimization. In evaluations on the large-displacement National Lung Screening Test (NLST) dataset, RWCNet exhibited superior performance (total registration error (TRE) of 2.11 mm) compared to other deep learning alternatives, and achieved results on par with variational optimization techniques. In contrast, on the OASIS dataset, which is characterized by smaller displacements, RWCNet achieved an average Dice similarity of 81.7%, representing only a modest improvement over other multi-scale deep learning models. Ablation experiments showed that multi-scale features consistently improved performance, whereas the correlation volume, number of recurrent steps, and inference-time instance optimization had large impacts on performance within the large-displacement NLST dataset. The performance of RWCNet compared to approaches that use instance optimization show that deep learning-based methods can find local minima that escape instance optimization methods. The results highlight the need for algorithm hyperparameter selection that adjusts with the dataset characteristics. RWCNet’s promising results may improve registration accuracy and computation efficiency, enabling many potential applications such as treatment planning, intra-procedural guidance, and longitudinal monitoring.
- 20 February 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (4) , 2084
Open AccessRecurrence with Correlation Network for Medical Image Registration
This work presents Recurrence with Correlation Network(RWCNet), a novel multi-scale recurrent neural network architecture for medical image registration that integrates core principles from optical flow, including correlation volume computation and inference-time instance optimization. In evaluations on the large-displacement National Lung Screening Test (NLST) dataset, RWCNet exhibited superior performance (total registration error (TRE) of 2.11 mm) compared to other deep learning alternatives, and achieved results on par with variational optimization techniques. In contrast, on the OASIS dataset, which is characterized by smaller displacements, RWCNet achieved an average Dice similarity of 81.7%, representing only a modest improvement over other multi-scale deep learning models. Ablation experiments showed that multi-scale features consistently improved performance, whereas the correlation volume, number of recurrent steps, and inference-time instance optimization had large impacts on performance within the large-displacement NLST dataset. The performance of RWCNet compared to approaches that use instance optimization show that deep learning-based methods can find local minima that escape instance optimization methods. The results highlight the need for algorithm hyperparameter selection that adjusts with the dataset characteristics. RWCNet’s promising results may improve registration accuracy and computation efficiency, enabling many potential applications such as treatment planning, intra-procedural guidance, and longitudinal monitoring.
- Open Access
- 30 January 2026
- Published by MDPI AG in Preprints.org
Recurrence with Correlation Network for Medical Image Registration
This work presents Recurrence with Correlation Network (RWCNet), a novel multi-scale recurrent neural network architecture for medical image registration that integrates core principles from optical flow, including correlation volume computation and inference-time instance optimization. In evaluations on the large-displacement National Lung Screening Test (NLST) dataset, which features large displacements, RWCNet exhibited superior performance (total registration error (TRE) of 2.11mm) to deep learning alternatives, and on par results with variational optimization techniques. In contrast, on the OASIS dataset characterized by smaller displacements, RWCNet's results (average dice similarity of 81.7\%) were superior to variational optimization techniques and showed a small improvement over other multi-scale deep learning models. Ablation experiments showed that multi-scale features consistently improved performance, where as the correlation volume, number of recurrent steps, and inference-time instance optimization only had large positive impacts on performance in the NLST dataset. The performance of RWCNet compared to approaches that use instance optimzation show that deep learning based methods can find local minima that escape instance optimization methods. The results highlight the need for algorithm hyperparameter selection that adjusts with the dataset characteristics. RWCNet's promising results may imporve registration performance and the speed of computation, allowing many potential applications including, treatment planning, intra-procedural guidance, and longitudinal monitoring.
- 30 January 2026
- Published by MDPI AG in Preprints.org
Open AccessRecurrence with Correlation Network for Medical Image Registration
This work presents Recurrence with Correlation Network (RWCNet), a novel multi-scale recurrent neural network architecture for medical image registration that integrates core principles from optical flow, including correlation volume computation and inference-time instance optimization. In evaluations on the large-displacement National Lung Screening Test (NLST) dataset, which features large displacements, RWCNet exhibited superior performance (total registration error (TRE) of 2.11mm) to deep learning alternatives, and on par results with variational optimization techniques. In contrast, on the OASIS dataset characterized by smaller displacements, RWCNet's results (average dice similarity of 81.7\%) were superior to variational optimization techniques and showed a small improvement over other multi-scale deep learning models. Ablation experiments showed that multi-scale features consistently improved performance, where as the correlation volume, number of recurrent steps, and inference-time instance optimization only had large positive impacts on performance in the NLST dataset. The performance of RWCNet compared to approaches that use instance optimzation show that deep learning based methods can find local minima that escape instance optimization methods. The results highlight the need for algorithm hyperparameter selection that adjusts with the dataset characteristics. RWCNet's promising results may imporve registration performance and the speed of computation, allowing many potential applications including, treatment planning, intra-procedural guidance, and longitudinal monitoring.
- 11 January 2026
- Published by Springer Nature in Annals of Biomedical Engineering
- Vol. 54 (5) , 1412-1424
Systemic Treatments Impact Bone Quality in a Preclinical Model of Mixed Femoral Metastases
Skeletal metastases disrupt bone remodeling, compromising mechanical integrity and increasing fracture risk. Cancer therapies can further influence bone quality. This study aimed to evaluate the impact of systemic cancer therapies (docetaxel (DTX) and zoledronic acid (ZA)) on bone quality in a preclinical model of mixed femoral metastases. Mixed metastases were induced in 5–6-week-old athymic male rats via intracardiac injection of luciferase-transfected ACE-1 canine prostate cancer cells (day 0). Healthy and inoculated rats were randomly assigned to receive no treatment, DTX (5 mg/kg), or ZA (60 μg/kg) on day 10. Tumor development was confirmed with bioluminescence imaging (day 14 and 21). Animals were euthanized on day 21. Bilateral femora were excised and underwent μCT scanning. Trabecular bone in the left femora was segmented for microstructural analysis. The left distal femora were then cut to 1 cm length and stained with barium sulfate for high-resolution μCT imaging to assess microdamage. The cut femora were then loaded to failure under axial compression. The right femora were processed for histology to evaluate bone histoarchitecture and verify tumor presence. DTX, despite lower bone mineral density, bone volume fraction, and trabecular number, showed reduced microdamage accumulation and improved load to failure in inoculated animals. ZA improved microstructural parameters, reduced damage volume fraction, and enhanced load to failure compared to inoculated untreated animals. These findings highlight the differential impact of cancer therapies on the quality of healthy and metastatic bone. Understanding these effects is essential for optimizing treatment strategies and minimizing skeletal complications secondary to skeletal metastases.
- 11 January 2026
- Published by Springer Nature in Annals of Biomedical Engineering
- Vol. 54 (5) , 1412-1424
Systemic Treatments Impact Bone Quality in a Preclinical Model of Mixed Femoral Metastases
Skeletal metastases disrupt bone remodeling, compromising mechanical integrity and increasing fracture risk. Cancer therapies can further influence bone quality. This study aimed to evaluate the impact of systemic cancer therapies (docetaxel (DTX) and zoledronic acid (ZA)) on bone quality in a preclinical model of mixed femoral metastases. Mixed metastases were induced in 5–6-week-old athymic male rats via intracardiac injection of luciferase-transfected ACE-1 canine prostate cancer cells (day 0). Healthy and inoculated rats were randomly assigned to receive no treatment, DTX (5 mg/kg), or ZA (60 μg/kg) on day 10. Tumor development was confirmed with bioluminescence imaging (day 14 and 21). Animals were euthanized on day 21. Bilateral femora were excised and underwent μCT scanning. Trabecular bone in the left femora was segmented for microstructural analysis. The left distal femora were then cut to 1 cm length and stained with barium sulfate for high-resolution μCT imaging to assess microdamage. The cut femora were then loaded to failure under axial compression. The right femora were processed for histology to evaluate bone histoarchitecture and verify tumor presence. DTX, despite lower bone mineral density, bone volume fraction, and trabecular number, showed reduced microdamage accumulation and improved load to failure in inoculated animals. ZA improved microstructural parameters, reduced damage volume fraction, and enhanced load to failure compared to inoculated untreated animals. These findings highlight the differential impact of cancer therapies on the quality of healthy and metastatic bone. Understanding these effects is essential for optimizing treatment strategies and minimizing skeletal complications secondary to skeletal metastases.
- Open Access
- 19 September 2025
- Published by Association for Computing Machinery (ACM)
- p. 16-21
Modality-Independent nnU-Net-Based Segmentation of Psoas Muscle and Vertebral Bodies in CT and MRI for Spinal Metastases Patients Undergoing SBRT
- 19 September 2025
- Published by Association for Computing Machinery (ACM)
- p. 16-21
Open AccessModality-Independent nnU-Net-Based Segmentation of Psoas Muscle and Vertebral Bodies in CT and MRI for Spinal Metastases Patients Undergoing SBRT
- Open Access
- 26 June 2025
- Published by Elsevier in Journal of the Mechanical Behavior of Biomedical Materials
- Vol. 170, 107116
Predicting rat lumbar vertebral failure patterns as synthetic μCT images using a deep convolutional generative adversarial network
- 26 June 2025
- Published by Elsevier in Journal of the Mechanical Behavior of Biomedical Materials
- Vol. 170, 107116
Open AccessPredicting rat lumbar vertebral failure patterns as synthetic μCT images using a deep convolutional generative adversarial network