Amir M. Vahdani

Amir M. Vahdani

Medical researcher | ML developer

Publications

You can also find my articles on my Google Scholar profile.

Journal Articles

U-ConvNext: A Robust Approach to Glioma Segmentation in Intraoperative Ultrasound

Published in Journal of Imaging Informatics in Medicine, 2025

This study is the result of an almost year-long effort on improving the SOTA on intraopreative ultrasound (iUS) segmentation. The end result is an improved U-net variant based on the ConvNext architecture leading to significant improvements in Dice score and Hausdorff 95th percentile distance, as well as a novel approach to uncertainty quantification (UQ) for semantic segmentation titled conformal segmentation. The choice to pursue this direction was motivated by the inherent uncertainty in iUS segmentation due to unclear borders and loss of fine detail in this modality.

Recommended citation: A.M. Vahdani, M. Rahmani, A. Pour-Rashidi, A. Ahmadian, P. Farnia, U-ConvNext: A Robust Approach to Glioma Segmentation in Intraoperative Ultrasound, Journal of Imaging Informatics in Medicine(2025). https://doi.org/10.1007/s10278-025-01648-7.
Download Paper

Towards trustworthy artificial intelligence in musculoskeletal medicine: A narrative review on uncertainty quantification

Published in Knee Surgery, Sports Traumatology, Arthroscopy, 2025

This paper could be considered the culmination of my studies on uncertainty quantification (UQ) and how it can be applied to biomedical ML problems. We introduce a very wide variety of UQ techniques (Bayesian, conformal prediction, evidential DL, test-time augmentation, etc) and propose a taxonomy that can help the reader better understand the overall landscape of these methods. We also showcase several studies that have successfully utilized these methods in musculoskeletal (MSK) medicine and focus on the translational potential of UQ, arguing for its necessity if DL models are to be utilized as reliable clinical tools. I was involved with all sections of the manuscript in this study and did a comprehensive literature review for this project; I also designed the graphical figures.

Recommended citation: A.M. Vahdani, M. Shariatnia, P. Rajpurkar, A. Pareek, Towards trustworthy artificial intelligence in musculoskeletal medicine: A narrative review on uncertainty quantification, Knee Surgery, Sports Traumatology, Arthroscopy (2025). https://doi.org/10.1002/ksa.12737.
Download Paper

Comparison of machine learning models with conventional statistical methods for prediction of percutaneous coronary intervention outcomes: a systematic review and meta-analysis

Published in BMC Cardiovascular Disorders, 2025

This paper compares a variety of supervised ML methods against logistic regression in predicting PCI outcomes, in terms of performance. We also extensively assess the models in terms of risk of bias (RoB) due factors such as data leakage and lack of external validation, using the CHARMS and PROBAST guidelines.

Recommended citation: S. Nayebirad, A. Hassanzadeh, A.M. Vahdani, A. Mohamadi, S. Forghani, A. Shafiee, F. Masoudkabir, Comparison of machine learning models with conventional statistical methods for prediction of percutaneous coronary intervention outcomes: a systematic review and meta-analysis, BMC Cardiovasc Disord 25 (2025) 310. https://doi.org/10.1186/s12872-025-04746-0.
Download Paper

Prescription patterns and the cost of antihyperglycemic drugs in patients with diabetes mellitus in Iran from 2014 to 2019

Published in Diabetes Research and Clinical Practice, 2025

This paper extensively analyzes prescription patterns for antihyperglycemic drugs in terms of factors such as physician specialist, time between follow-up sessions, comorbidities, etc., using large scale data directly from Iran Health Insurance (IHI) servers. I was the primary data analyst in this project, performing all steps of data retrieval, cleaning and processing. The analysis included techniques such as GESD and SHAP value usage for intrepreting factors affecting overall prescription costs.

Recommended citation: [1] M. Rezaee, M.M. Nasehi, Z. Aminzade, H. Karami, A.M. Vahdani, R. Daroudi, M. Effatpanah, L. Ghamkhar, M. Heidari-Foroozan, M. Arab, Z. Shahali, R. Mehrizi,Prescription patterns and the cost of antihyperglycemic drugs in patients with diabetes mellitus in Iran from 2014 to 2019, Diabetes Research and Clinical Practice (2025) 112078. https://doi.org/10.1016/j.diabres.2025.112078.
Download Paper

Deep Conformal Supervision: Leveraging Intermediate Features for Robust Uncertainty Quantification

Published in Journal of Imaging Informatics in Medicine, 2024

This paper is about improving the coverage and efficiency of conformal prediction for deep neural networks, using novel non-conformity score computation methods. Specifically, the focus is on improving the robustness of medical image classification models (CNNs) via improved uncertainty quantification techniques.

Recommended citation: A.M. Vahdani , S. Faghani, Deep Conformal Supervision: Leveraging Intermediate Features for Robust Uncertainty Quantification, J Digit Imaging. Inform. Med. (2024). https://doi.org/10.1007/s10278-024-01286-5.
Download Paper

Deep Learning‐Based Techniques in Glioma Brain Tumor Segmentation Using Multi‐Parametric MRI: A Review on Clinical Applications and Future Outlooks

Published in Journal of Magnetic Resonance Imaging, 2024

This paper is a comprehensive review of different approaches for glioma MRI segmentation, cinluding non-deep learning methods. We provide an overall narrative of how this field has evolved (up to and including how transformer architectures are increasingly used here), and mention key studies such as the nnU-net model, among others, diving into key innovations and the reasoning behind a variety o methodologies; we also discuss relevant datasets (e.g., different BraTS versions). I was a co-first author in this work, and wrote the sections focusing on different models and architectures, reading a lot of the literature on glioma segmentation during the literature synthesis and manuscript writing process.

Recommended citation: D.J. Ghadimi, A.M. Vahdani, H. Karimi, P. Ebrahimi, M. Fathi, F. Moodi, A. Habibzadeh, F. Khodadadi Shoushtari, G. Valizadeh, H. Mobarak Salari, Deep Learning‐Based Techniques in Glioma Brain Tumor Segmentation Using Multi‐Parametric MRI: A Review on Clinical Applications and Future Outlooks, Journal of Magnetic Resonance Imaging (2024). https://doi.org/10.1002/jmri.29543.
Download Paper

Epidemiological Insights into Diabetic Foot Amputation and its Correlates: A Provincial Study

Published in Clinical Medicine Insights: Endocrinology and Diabetes, 2024

This paper was one of my first research experiences. My main role was in writing the manuscript, start to finish.

Recommended citation: M. Aalaa, A.M. Vahdani, M. Mohajeri Tehrani, N. Mehrdad, M. Zohdirad, M. Sadati, M. Amini, S. Mehrpour, M. Ebrahimi, B. Larijani, Epidemiological Insights into Diabetic Foot Amputation and its Correlates: A Provincial Study, Clinical Medicine Insights: Endocrinology and Diabetes 17 (2024) 11795514241227618. https://doi.org/10.1177/11795514241227618.
Download Paper

Assessment of quality of life and its affecting factors in osteosarcopenic individuals in the Iranian older adult population: Bushehr Elderly Health (BEH) program

Published in Osteoporosis and Sarcopenia, 2023

This paper was one of my first research experiences. My main role was in writing the manuscript, start to finish.

Recommended citation: A.M. Vahdani, M. Sanjari, N. Fahimfar, M. Ebrahimpur, G. Shafiee, K. Khalagi, M.J. Mansourzadeh, I. Nabipour, B. Larijani, A. Ostovar, Assessment of quality of life and its affecting factors in osteosarcopenic individuals in the Iranian older adult population: Bushehr Elderly Health (BEH) program, Osteoporosis and Sarcopenia 9 (2023) 142–149.
Download Paper