Skip to content
View ThalesMMS's full-sized avatar

Block or report ThalesMMS

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.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

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

Report abuse
ThalesMMS/README.md

Thales Matheus M. Santos

Physician–Radiologist · Computer Science (B.Sc. in progress)

Email LinkedIn


About Me

I'm a physician and computer science undergraduate exploring how software engineering and artificial intelligence can advance radiological practice. My radiology experience is hospital-based diagnostic imaging, including emergency/trauma and oncologic care. I have hands-on contributions to DICOM viewers, PACS infrastructure, and exploratory AI tooling for healthcare and medical imaging, and I'm comfortable moving from clinical questions to code and back. My main technical interests include:

  • DICOM viewers with 2D/MPR/volumetric rendering and DIMSE/DICOMweb integration
  • Generative AI in medicine (LLMs/Agents, local inference, RAG/finetuning)
  • Infrastructure for healthcare systems (PACS, NAS, EHR, and other systems)
  • AI-assisted software engineering and developer tooling
  • Full-stack development tailored to clinical workflows

Interests

  • Local-first imaging platforms and DICOM viewers
  • AI applications in medicine
  • PACS servers in low-resource environments
  • AI-assisted radiological decision-making

Current Projects

Isis DICOM Viewer (Windows, macOS, iOS, iPadOS): built natively in Swift for Apple platforms and developed in Qt/C++ for Windows, providing 2D visualization, MPR, volumetric rendering, PACS/DIMSE support, and ROI measurement tools (closed source).

JFlutter: Cross-platform interactive educational tool to design and simulate automata, regular expressions, formal languages and Turing machines. Mobile‑first, touch‑optimised UI.

Dicom-Tools: Single UI to test major DICOM libraries across multiple languages (DCMTK, GDCM, ITK/VTK, dicom-rs, fo-dicom, dcm4che, Cornerstone3D, pydicom/pynetdicom, etc.).

PACS-Natural-Language-Query: toolkit for querying PACS studies using natural language.

MTK - Metal Toolkit: Modern Swift/Metal toolkit for high-fidelity medical imaging (volume rendering, MPR, SceneKit integration, SwiftUI components).

Other Projects

DICOM-Decoder: Swift DICOM pixel decode and window/level pipeline for uncompressed images.

Radiology-Templates: Radiology report templates and Rust/Python tools to convert across DOCX, Markdown, and TXT.

DcmSwift: Swift-native DICOM core (DIMSE + DICOMweb).

TotalSegmentator Horos/Osirix Plugin: Bringing the modern TotalSegmentator to the open-source Horos Project!

GitHub-Replicant: fast async Rust CLI for bulk GitHub backups (clone/pull, sync by starred/followers/following).

Code-Scanner: generate clean “whole-repo” text snapshots for reviews, audits, and AI prompts.

Orthanc‑PACS: Turn‑key Orthanc Docker stack with SQLite, Orthanc Explorer 2, DICOMweb, OHIF viewer, key plugins and sensible defaults.

mammography-pipelines-py: Mammography experimentation repo delivering reproducible preprocessing, feature engineering, and modeling steps. Run ResNet50 or EfficientNet B0 extractors for full training loops and density classifiers.

brain‑mri‑pipelines‑py: Experimental Alzheimer’s detection framework using MRI + clinical fusion (research playground).

Lung‑Nodule‑app: Guideline calculators for Fleischner 2017 and ACR Lung‑RADS v2022. Educational use only.

Rust‑Neural‑Networks and Swift‑Neural‑Networks: Small neural network models in Rust and Swift for MNIST digit classification and XOR tasks. Includes MLP, CNN and single‑head self‑attention architectures, along with Python utilities for visualization and digit recognition.

reports‑to‑llm: Rust tool that converts DOCX/RTF medical reports into clean, UTF‑8‑encoded text optimized for LLM training.

Horos‑Backup‑Script: Python script and macOS LaunchAgent to automate bulk backups of CT/MR studies from Horos.

OsiriX‑Backup‑Plugin: Swift plugin for OsiriX/Horos that automates sending DICOM studies to remote PACS.

WALL-ET: Mobile Bitcoin wallet (WIP).

jff‑to‑tex‑Turing‑Machine‑Diagram‑Converter: Python utility that converts JFLAP Turing machine diagrams into LaTeX/TikZ diagrams.

LaTeX‑Paper‑Template: Ready‑to‑use LaTeX template for SBC articles/monographs with modular chapters, placeholder content and centralized metadata.

Orthanc for QNAP: Custom .qpkg packaging


Tech stack

  • Languages: Python · Swift · C++ · Rust · Dart · JavaScript · Java
  • Frameworks: SwiftUI · Django · Flutter · Qt · Metal
  • AI/ML: PyTorch · TensorFlow · scikit-learn
  • Medical imaging: DCMTK · GDCM · VTK · ITK · OsiriX/Horos · 3D Slicer · Orthanc
  • Infra/DevOps: Docker · Proxmox · NAS (QNAP + OSS) · Tailscale · Git
  • Environments: macOS + Windows · Linux VMs · Dockerized services

Top languages

Pinned Loading

  1. TotalSegmentator-OsiriX-Horos-Plugin TotalSegmentator-OsiriX-Horos-Plugin Public

    Prototype OsiriX/Horos plugin that exports the active series, runs TotalSegmentator inside its managed Python environment, and reimports the labeled results.

    Python 1

  2. MTK MTK Public

    MTK (Metal Toolkit) is a modern Swift/Metal toolkit for medical imaging, delivering volume rendering, MPR, projections and reusable components.

    Swift 1

  3. JFlutter JFlutter Public

    Cross-platform educational interactive tool to design and simulate automata, formal languages, regular expressions and Turing Machines.

    Dart 3

  4. Dicom-Tools Dicom-Tools Public

    Unified UI to test DCMTK, GDCM, ITK, VTK, dicom-rs, fo-dicom, dcm4che, Cornerstone3D, pydicom, pynetdicom, python-gdcm, SimpleITK and dicom-numpy.

    Python 1

  5. DICOM-Decoder DICOM-Decoder Public

    Pure swift library delivering DICOM decoding core that reads uncompressed studies, exposes metadata and pixel buffers and applies window/level adjustments.

    Swift 1

  6. PACS-Natural-Language-Query PACS-Natural-Language-Query Public

    DICOM PACS toolkit for querying studies using natural language.

    Python 1