Skip to content
View hash123shaikh's full-sized avatar
🎯
Focusing
🎯
Focusing
4 followers · 24 following

Block or report hash123shaikh

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
hash123shaikh/README.md

Hasan Shaikh. I build AI that can tell, before treatment, how a head and neck cancer will behave.

Portfolio Google Scholar ORCID LinkedIn Email CV

Who is this?

I teach machines to read cancer scans, and then I check whether they are telling the truth.

I am a data scientist in the Quantitative Imaging Research and AI Lab (QIRAIL), Department of Radiation Oncology, Christian Medical College, Vellore, one of India's leading cancer centres. My work asks one question over and over: before we treat, can we tell who is likely to relapse?

I built a radiomics signature that predicts head and neck cancer recurrence (AUC 0.82), a segmentation model that outlines tumours from CT alone, and the data pipelines behind a 2,000+ patient imaging study. Then I reproduced a published AI model with the authors' own code and showed that its headline numbers did not hold up. Building models and breaking them: both are the job.

I am looking for a PhD position in medical imaging and AI for oncology

How my work runs

Five steps: scan, outline, measure, select, predict

A planning CT goes in. A 3D nnU-Net outlines the tumour. PyRadiomics turns the outline into more than 1,000 numbers about shape, brightness and texture. Metaheuristic search keeps the few that are stable and useful. A model then estimates recurrence risk before treatment, with SHAP showing a clinician why.

What I work on

  • Radiomics-based outcome prediction for head and neck cancer
  • Automated tumour segmentation with 3D CNNs and nnU-Net
  • Adaptive radiotherapy and treatment-planning automation
  • Uncertainty quantification in clinical AI
  • Reproducibility and trustworthy AI: CLAIM and TRIPOD+AI assessments
  • Multimodal deep learning across imaging, clinical and genomic data
  • DICOM/PACS engineering and clinical data pipelines (Orthanc, XNAT, AWS S3)

Publications

# Work Status
1 Anatomy-guided skip-connection recalibration for CT-based head and neck tumor auto-segmentation. International Journal of Biomedical Imaging Under submission
2 Can reproducibility be achieved? External validation of "reproducible" image-based prognostic models for head and neck cancer. Scientific Reports Under submission
3 Development and validation of a prospective radiomics-clinical signature for locoregional recurrence in locally advanced head and neck cancer. ESTRO 2026, Stockholm Poster highlight (abstract)
4 Metaheuristic-driven machine learning pipelines for radiomics-based prediction of locoregional recurrence in head and neck cancer. AIHC, Springer LNEE, 2025 Accepted (PDF)
5 Automated segmentation of head and neck cancer from CT images using 3D convolutional neural networks. medRxiv, 2026 Preprint
6 Cancer survival prediction using artificial intelligence: current status and future prospects. Data Science in the Medical Field, Elsevier, 2024 Book chapter

Full list on Google Scholar and ORCID.

Talks and posters

  1. Poster, 14th Research Day, CMC Vellore, October 2024: Can CT radiomics predict recurrence in head and neck cancer? Early results from a prospective imaging trial.
  2. Oral, third prize, 2nd National Symposium on Health Data and AI, CMC Vellore, March 2025: Automated segmentation of head and neck cancer from CT images using 3D nnU-Net.
  3. Talk, 15th Research Day, CMC Vellore, October 2025: Before we treat, can we tell? A locoregional recurrence signature in head and neck cancer. (slides)
  4. Paper, AIHC 2025, NIT Calicut, December 2025: Metaheuristic-driven radiomics for recurrence prediction.
  5. Poster highlight, ESTRO 2026, Stockholm, Sweden.

Building the community

  • Workshop on Radiomics and Auto Segmentation: I organised all three editions at CMC Vellore (14 to 15 November 2025, 27 to 28 March 2026, 18 to 19 September 2026), hands-on training for clinicians, medical physicists and researchers. Certificate for the first edition.
  • Head and Neck Cancer Survivorship Program: organised at CMC Vellore, 12 to 13 September 2025.
  • 2nd Annual Winter Symposium on Health Data and AI: organising team, March 2025.

The path

Timeline of education and work

Toolbox

Python PyRadiomics nnU-Net MedSAM TensorFlow scikit-learn Pandas Docker PostgreSQL Git

Imaging: 3D Slicer · ITK-SNAP · Orthanc PACS · XNAT · DICOM · NIfTI
ML: TensorFlow · Keras · scikit-learn · PyRadiomics · nnU-Net · SHAP
Methods: radiomics · metaheuristic optimisation · graph neural networks
Infra: Docker · GitHub · PostgreSQL · AWS S3 · Bash

What I believe

  • AI will eventually automate radiation oncology end to end: delivery, planning, auto-segmentation, adaptive radiotherapy and a paperless, AI-driven medical record. We are close.
  • The biggest impact of medical AI will not be in well-resourced hospitals. It will be where expert clinicians are scarce.

Let's talk

I like talking with researchers, clinicians, engineers and anyone curious about AI in healthcare. Research, PhD opportunities, collaborations: all welcome.

📍 Vellore, Tamil Nadu, India · ✉️ hasanshaikh3198@gmail.com · 🌐 hash123shaikh.github.io

Making radiotherapy smarter, one pipeline at a time.

Pinned Loading

  1. cervical-cancer-ml-pipeline cervical-cancer-ml-pipeline Public

    End-to-end ML pipeline for cervical cancer risk-based screening stratification

    Python

  2. reproducibility-hnc-cnn reproducibility-hnc-cnn Public

    Independent reproduction and external validation of image-based deep learning prognostic models for head and neck cancer (Mateus et al., Sci Rep 2023): exact & statistical reproducibility, external…

    Python

  3. Master-Thesis Master-Thesis Public

    Multimodal deep learning for cancer survival prediction (91.2% accuracy, 79.8% sensitivity) using Gated Attention CNNs + Random Forest

    Python

  4. how-to-remove-sensitive-data-from-github how-to-remove-sensitive-data-from-github Public

    A perfect guide on how to prevent and remove `.env` files or sensitive data from GitHUb repositries. Includes cleanup commands, best practics, and sample files for safe code sharing.