https://ramtin-mojtahedi.github.io/#news Verified research updates and original public LinkedIn posts from Ramtin Mojtahedi. en-ca Wed, 04 Nov 2026 12:47:03 +0000 1440 daily 1 https://www.linkedin.com/feed/update/urn:li:activity:7518452739480215552/ Registration is open for the Cross-Institute Research Mentorship Program across University Health Network. I helped design the program as part of UHN Postdoc Association (UHNPA), working with UHN Office of Research Trainees. It connects trainees with mentors outside their own lab for career advice, support, and connections across UHN. Program dates: January–May 2027 Time commitment: At least one meeting per month, virtually or in person. • Mentees: UHN graduate students and postdoctoral researchers. Register as a mentee: https://lnkd.in/gEBigcbu • Mentors: Experienced UHN postdocs, scientists, scientific associates, PIs, clinician-scientists/investigators, and research staff. Register as a mentor: https://lnkd.in/gpaeZ3-4 If you’d like guidance beyond your own lab or have experience to share, we’d love to have you involved. Certificates are available upon completing the program requirements. Full program details: https://lnkd.in/gNFSYShJ Please share with UHN colleagues who may be interested. Research at UHN University of Toronto Temerty Faculty of Medicine, University of Toronto Toronto Metropolitan University #UHN #UHNPA #Mentorship #CareerDevelopment #GraduateStudents #Postdocs #UofT #TemertyMed #TMU LinkedIn post · Mentorship ramtin-news:linkedin-activity-7518452739480215552 Tue, 20 Oct 2026 00:00:00 +0000 Original image attached to Ramtin Mojtahedi’s LinkedIn post about the Cross-Institute Research Mentorship Program Ramtin Mojtahedi · Original LinkedIn post media https://www.linkedin.com/feed/update/urn:li:activity:7518443940417601536/ Celebrating 100 years of Homecoming at Queen's University last weekend was a reminder of the value of staying connected across generations and disciplines. My PhD years at Queen's School of Computing, within Queen's University Faculty of Arts and Science, shaped my work in AI and my commitment to teaching and mentorship. For me, alumni connections offer a practical way to contribute: sharing experience, exchanging ideas and helping students navigate their next steps. I’m always glad to connect with students and fellow alumni interested in AI, research or mentorship. Thank you to the Queen's Alumni team, organizers and volunteers for a memorable weekend. #QueensU #QueensAlumni #QueensComputing #QueensArtsAndScience #Homecoming2026 LinkedIn post · Alumni ramtin-news:linkedin-activity-7518443940417601536 Tue, 20 Oct 2026 00:00:00 +0000 Original image attached to Ramtin Mojtahedi’s LinkedIn post about Queen’s Homecoming Ramtin Mojtahedi · Original LinkedIn post media https://www.linkedin.com/feed/update/urn:li:activity:7516964444380487680/ The value of AI in biomedical research depends on how well computational methods address the underlying scientific question. Attending the 2026 iBEST | Institute for Biomedical Engineering, Science and Technology Symposium, bringing together Toronto Metropolitan University and St. Michael’s Hospital at Unity Health Toronto, was an opportunity to explore research beyond my own area and consider where complementary expertise could lead to collaboration. As an AI researcher, I’m interested in translating biomedical questions into rigorous computational studies. That involves understanding the limitations of the data, defining the insight we hope to gain and designing validation around the intended application. Agreeing early on what would constitute meaningful evidence gives a collaboration a clear direction. It guides methodological choices and keeps the work connected to the scientific problem that motivated it. #ArtificialIntelligence #BiomedicalResearch #ResearchCollaboration #InterdisciplinaryResearch LinkedIn post · Biomedical AI ramtin-news:linkedin-activity-7516964444380487680 Fri, 16 Oct 2026 00:00:00 +0000 Original image attached to Ramtin Mojtahedi’s LinkedIn post about the 2026 iBEST Symposium Ramtin Mojtahedi · Original LinkedIn post media https://arxiv.org/abs/2610.04151 Our new public preprint benchmarks invisible watermarks and content fingerprints as C2PA soft bindings across images, audio and video. The openly licensed benchmark includes reproducible code and recorded results. Preprint; not peer reviewed. Preprint · Content provenance ramtin-news:soft-bindings-benchmark-preprint-20261002 Fri, 02 Oct 2026 00:00:00 +0000 Preprint figure comparing watermark and fingerprint success across image, audio and video transformations Kazempourradi, Mojtahedi & Mohseni · Preprint figure https://arxiv.org/abs/2609.14106 A public manuscript of our SPIE 2026 paper on parameter-efficient fine-tuning of Segment Anything for liver-tumour segmentation is now available on arXiv. It compares LoRA, QLoRA, a convolutional adapter and DiSCo across prompt settings and compute measures. This arXiv manuscript is not a separate peer-reviewed publication. Public manuscript · Medical imaging ramtin-news:peft-sam-liver-ct-manuscript-20260912 Sat, 12 Sep 2026 00:00:00 +0000 https://pycad.co/blog/point-box-colorectal-liver-met-overlay-ct/ PYCAD covered our new DiSECT and SiGA preprint, discussing spectral adaptation of Segment Anything for colorectal liver metastasis segmentation. This is research coverage, not a clinical product announcement. Research coverage ramtin-news:pycad-spectral-adapters-coverage-20260911 Fri, 11 Sep 2026 00:00:00 +0000 GitHub profile image for Ramtin Mojtahedi GitHub profile image https://arxiv.org/abs/2609.11703 Our DiSECT and SiGA adapters adapt Segment Anything for colorectal liver metastasis segmentation. In the 91-case held-out, no-prompt evaluation, SiGA achieved 0.76 Dice on tumour-positive CT slices. Preprint; not peer reviewed. Preprint ramtin-news:spectral-adapters-preprint-2026 Thu, 10 Sep 2026 00:00:00 +0000 Published diagram showing CT images, SAM encoders, adapters and tumour segmentation masks Mojtahedi et al. · Preprint figure 1 https://www.umbjournal.org/article/S0301-5629(26)00197-3/fulltext Our co-first-author study appears in Ultrasound in Medicine & Biology, volume 52, issue 9, pages 1890–1899. The work combines carotid plaque radiomic features with machine learning for cardiovascular risk prediction. Journal issue ramtin-news:carotid-plaque-september-issue-2026 GitHub profile image for Ramtin Mojtahedi GitHub profile image https://www.nature.com/articles/s41598-026-60860-9 My co-first-author study combines clinical and CT-derived imaging data for preoperative risk prediction in colorectal liver metastases. Springer Nature published the version of record on 14 October 2026. Publication ramtin-news:scientific-reports-multimodal-2026 Tue, 14 Jul 2026 00:00:00 +0000 GitHub profile image for Ramtin Mojtahedi GitHub profile image