Course Objective & Scope
Medical images (CT, MRI, X-ray, ultrasound, digital pathology) differ fundamentally from standard 8-bit photos: 16-bit Hounsfield units, anisotropic 3D voxel spacings, gigapixel slides and patient-level data leakage require specialised tooling. Starting from data preparation with PyDICOM and SimpleITK, this programme covers 2D and 3D medical image classification and segmentation in depth using MONAI and nnU-Net.
Organ and lesion volumetry in radiology, gigapixel Whole Slide Image (WSI) patch analysis in pathology, sensitivity/specificity calibration and PACS integration are practised hands-on. Participants build a 3D segmentation and clinical evaluation pipeline on a medical dataset.
Who Should Attend?
- AI engineers working with radiology, pathology, ophthalmology or cardiology imaging
- Data scientists developing medical devices, Software as a Medical Device (SaMD) and bioinformatics tools
- Researchers applying deep learning in academic and clinical imaging studies
- Prerequisite: PyTorch and prior experience with deep learning for images
Key Takeaways
- Prepare DICOM and NIfTI volumes through HU windowing, isotropic resampling and de-identification
- Train organ and lesion segmentation models on CT and MRI volumes with MONAI, 3D U-Net and nnU-Net
- Prevent data leakage via patient-level splits and compute Dice and Hausdorff distance metrics
- Apply patch extraction and Multiple Instance Learning on gigapixel digital pathology (WSI) slides
- Prepare model outputs for DICOM-SEG / PACS workflows and SaMD clinical validation requirements
Curriculum
Medical Image Formats and Preprocessing
- DICOM tags, de-identification, NIfTI and anatomical coordinate systems
- Hounsfield Unit (HU) windowing, isotropic voxel resampling and intensity normalisation
3D Volumetric Segmentation with MONAI and nnU-Net
- 3D U-Net, Swin UNETR and the self-configuring nnU-Net framework
- 3D patch sampling, memory management and combined Dice + Cross-Entropy loss
Radiology Classification, Detection and Foundation Models
- Multi-label classification and lesion localisation in chest X-rays and mammography
- Adapting MedSAM and BiomedCLIP to sparsely labelled medical datasets
Digital Pathology and Whole Slide Image (WSI) Analysis
- Reading gigapixel slides with OpenSlide, tissue masking and multi-scale patch extraction
- Multiple Instance Learning (MIL) on weakly labelled biopsy slides
Clinical Evaluation, Calibration and PACS Integration
- Sensitivity/specificity, ROC/FROC, uncertainty estimation and cross-scanner domain shift
- Generating DICOM-SEG outputs, PACS/RIS integration and SaMD (MDR/FDA) validation principles
Other Courses in This Category
-
Reception and lobby queue/crowd analytics, buffet tray replenishment and food-waste measurement, car-park ANPR, pool/facility safety and privacy-compliant anonymous analytics, for hospitality technology teams.
- Live Course
- 5 modules
-
Model conversion with ONNX Runtime, TensorRT, OpenVINO and TFLite, INT8 quantization, camera pipelines on NVIDIA Jetson and Raspberry Pi, GStreamer and OTA updates, for teams running vision models at low latency in the field.
- Live Course
- 5 modules
-
Knowledge distillation, structured pruning, Quantization-Aware Training (QAT), NAS and Torch-TensorRT profiling, for engineers shrinking and accelerating large computer vision models with minimal accuracy loss.
- Live Course
- 5 modules
-
Computer Vision
Comprehensive Computer Vision training ranging from classical image processing techniques to PyTorch-based CNN, Vision Transformer (ViT), Swin Transformer, and ConvNeXt architectures. Hands-on coverage includes object detection with YOLO and DETR, segmentation with SAM and U-Net, video analysis, 3D vision (NeRF, Gaussian Splatting), Diffusion models, and model deployment with ONNX/TensorRT.
- Live Course
- 13 modules
-
Digital image basics, filtering and feature extraction with OpenCV, the classification/detection/segmentation task types and a first prototype with pretrained models, for technical teams scoping computer vision use cases.
- Live Course
- 5 modules
-
CNN architectures (ResNet, EfficientNet, ConvNeXt) in PyTorch, transfer learning, data augmentation and training loops, for developers and data scientists who need to train and evaluate image models on company data.
- Live Course
- 5 modules
Get in touch
