NeuroSeg Pro is an enterprise-grade clinical desktop suite engineered for rapid, automated 3D multi-modal MRI brain tumor segmentation. Leveraging state-of-the-art MONAI DynUNet architectures and hierarchical sigmoid post-processing, NeuroSeg Pro isolates and characterizes sub-tumoral regions with sub-voxel accuracy.
📥 Download Windows Installer (v4.0.0) • 📖 Read Documentation • 🐛 Report Issue
Version 4.0.0 introduces a redesigned, zero-friction Windows deployment pipeline and deep architectural optimizations:
-
⚡ Zero-Touch Automated Bootstrapper: The installer automatically audits system health, acquires Python 3.11 silently if missing, provisions isolated
%LOCALAPPDATA%\NeuroSegPro\.venvvirtual environments, and installs required clinical deep learning toolchains. -
📦 Intelligent Package Reuse: Virtual environments utilize
--system-site-packages, eliminating multi-gigabyte PyTorch/MONAI re-downloads on workstations where libraries or CUDA runtimes already exist. -
🔕 Windowless Silent Execution: Integrated VBS launcher (
NeuroSegPro.vbs) launches the PyQt5 desktop suite cleanly without console command prompt windows or clutter. -
🩺 Self-Healing Repair Tool: Built-in automated diagnostic suite detects corrupted environments or deleted runtimes and provides instant GUI repair prompts (
install.ps1 -Repair). -
🔬 Accurate Volumetric Inference: Fixed 6-stage DynUNet spatial bottleneck strides (
$S/64 \rightarrow S/32 \rightarrow S/16 \rightarrow S/8 \rightarrow S/4 \rightarrow S/2 \rightarrow S$ ), enabling seamless 3D sliding-window segmentation on large NIfTI scans (roi_size=128³).
NeuroSeg Pro processes standard multi-parametric MRI sequences (T1, T1ce, T2, FLAIR) in NIfTI format (.nii / .nii.gz) and outputs hierarchical tumor sub-region masks adhering to international clinical standards:
graph LR
A["T1 scan"] --> E["Multi-Modal 4D Stack"]
B["T1ce scan"] --> E
C["T2 scan"] --> E
D["FLAIR scan"] --> E
E --> F["MONAI DynUNet Engine"]
F --> G["Sliding-Window Inference (128x128x128 / overlap=0.5)"]
G --> H["Hierarchical Sigmoid Thresholding"]
H --> I["Label 3: Enhancing Tumor (ET)"]
H --> J["Label 1: Tumor Core (TC / NCR)"]
H --> K["Label 2: Whole Tumor (WT / ED)"]
- Label 3 (Enhancing Tumor - ET): High-priority hyper-intense regions on T1ce (overwrites all underlying labels).
- Label 1 (Non-Enhancing / Necrotic Core - NCR/NET): Core necrotic structures within the tumor margin.
- Label 2 (Peritumoral Edema / Whole Tumor - ED/WT): Surrounding FLAIR hyper-intense infiltrative edema.
| Component | Minimum Specification | Recommended Clinical Workstation |
|---|---|---|
| Operating System | Windows 10 (64-bit) | Windows 11 Pro / Enterprise (64-bit) |
| Processor | Intel Core i5 / AMD Ryzen 5 (4 Cores) | Intel Core i7/i9 / AMD Ryzen 7/9 (8+ Cores) |
| System Memory (RAM) | 16 GB RAM | 32 GB+ RAM (for multi-modality 3D loading) |
| Graphics Card (GPU) | CPU Fallback Mode (No GPU required) | NVIDIA GeForce RTX 3060 / 4070 / RTX A2000+ (8 GB+ VRAM) |
| CUDA Toolchain | Not Required for CPU Mode | CUDA 11.8+ / 12.1+ Runtime Driver |
| Storage Space | 6.0 GB Free Disk Space | NVMe SSD with 10 GB+ Free Space |
Download the compiled standalone Windows installer from our official GitHub Release:
👉 NeuroSegPro_Setup_v4.0.0.exe (2.89 MB)
- Double-click
NeuroSegPro_Setup_v4.0.0.exe. - Accept the standard Windows User Account Control (UAC) Administrator prompt.
- Choose your desktop icon preferences and click Install.
Upon completion, the installer automatically triggers install.ps1, which orchestrates your local clinical runtime:
- Creates
%LOCALAPPDATA%\NeuroSegPro\.venv. - Acquires PyTorch (
torch-2.5.1+cu121), MONAI, Nibabel, SciPy, and PyQt5. - Registers
.niiand.nii.gzfile associations so you can open study files directly from Windows Explorer.
Launch the application via your Desktop Shortcut or Start Menu!
C:\Program Files\NeuroSeg-Pro\ <-- Core Application Bundle
│
├── NeuroSegPro.vbs <-- Windowless Silent VBS Launcher
├── install.ps1 <-- Intelligent Runtime Bootstrapper
├── app\ <-- Python Application Source Suite
│ ├── main.py <-- Application Entrypoint
│ ├── core\
│ │ ├── custom_model.py <-- DynUNet & ParallelQuantumBottleneck
│ │ ├── inference.py <-- Sliding-Window Clinical Engine
│ │ └── loader.py <-- NIfTI Multi-Modal Reader
│ └── ui\ <-- PyQt5 Themeable Interface Components
└── assets\ <-- Icons and Visual Brand Styling
%LOCALAPPDATA%\NeuroSegPro\ <-- Isolated Runtime Environment
│
├── .venv\ <-- Dedicated Python 3.11 Virtual Environment
└── logs\ <-- Multi-Log Split Diagnostic Telemetry
├── bootstrap.log <-- Lifecycle & OS Audit Log
├── python_install.log <-- Silent Acquisition Log
└── gpu_audit.log <-- CUDA & PyTorch Engine Verification
Why did setup display "Runtime components have not yet been installed"?
If your network connection dropped or antivirus blocked setup, simply double-click the Repair & Maintenance Tool shortcut in your Start Menu or run:
powershell -ExecutionPolicy Bypass -File "C:\Program Files\NeuroSeg-Pro\install.ps1" -Repair
Where can I inspect log files if installation fails?
All diagnostic telemetry is persisted in plain-text inside your local profile:
C:\Users\<YourUsername>\AppData\Local\NeuroSegPro\logs\bootstrap.log
Does NeuroSeg Pro work on machines without an NVIDIA GPU?
Yes! The inference engine automatically detects hardware capabilities. If CUDA is unavailable, it gracefully drops back to multi-core CPU sliding-window processing.
If you wish to contribute to source code or build custom installers locally:
- Clone Repository:
git clone https://github.com/mohany203/NeuroSeg-Pro.git cd NeuroSeg-Pro
- Stage Release & Compile Installer:
Ensure Inno Setup 6 is installed, then execute:
The compiled production installer will be output to
.\stage_release.ps1 & "C:\Program Files (x86)\Inno Setup 6\ISCC.exe" installer.iss
dist\NeuroSegPro_Setup_v4.0.0.exe.
 Mohamed Hany Application Architect & GUI Suite |
 Omar Eldash AI Model & Volumetric Inference |
 Ahmed Samy Deep Learning & AI Training |
Distributed under the MIT License. See LICENSE for detailed terms.