CNN-GAN Cancer Image Classifier

$30-250 USD

Posted 1 minute ago

•

Ends in 6 days

$30-250 USD

Paid on delivery

I have four independent, pre-labelled imaging datasets—MRI, ultrasound, mammogram and histology—each split into the two diagnostic classes that matter most to clinicians: benign and malignant. My goal is to write a single, reusable codebase that I can launch against any one of these datasets and obtain the highest possible classification accuracy. The pipeline must combine a convolutional backbone for feature extraction, a GAN-based data-augmentation module to balance classes and enlarge the training set, and Grad-CAM to visualise the discriminative regions that drive each prediction. Because the datasets differ in resolution and appearance, the code should accept modality-specific preprocessing parameters but keep the core architecture, training loop and evaluation logic identical; I will simply point the script to the chosen dataset and let it run. Key expectations • End-to-end Python code (TensorFlow or PyTorch) that trains, validates and tests the model, outputs accuracy, AUC and confusion matrix, and saves Grad-CAM heat-maps alongside original images. • Clear configuration section so I can switch between the four datasets without touching the model code itself. • Well-commented notebook or script plus a short README explaining environment setup and how to reproduce your results. Deliverables will be considered complete when the code executes successfully on at least one sample from each modality and the Grad-CAM overlay correctly highlights tumour regions. If this challenge matches your skill set in CNNs, GANs and medical-image explainability, let’s collaborate.

Project ID: 40812795

About the project

18 proposals

Open for bidding

Remote project

Active 9 secs ago

Place your bid

Bid amount Email address

Benefits of bidding on Freelancer

Set your budget and timeframe

It's free to sign up and bid on jobs

18 freelancers are bidding on average $128 USD for this job

User Avatar
Flag of INDIA

Hello, I can help you with "CNN-GAN Cancer Image Classifier" as per your given set of instructions. Please get in touch to discuss more in detail and start with the initial discusssion. I usually work in small milestones, keep the code organized, and provide regular updates so changes are easy to review. Can you share a few more details? I'll suggest the best implementation and an accurate timeline. ⭐ 5.0/5 from a recent client: "Excellent work. Would definitely utilize again. Stellar job!" Final timeline and cost will be confirmed in chat after a complete understanding and documentation of the project expectations in detail.

User Avatar
Flag of TANZANIA, UNITED REPUBLIC OF

Hello, Looking at your four modalities (MRI, ultrasound, mammogram, histology) all sharing the same benign vs malignant task, the cleanest path is a config-driven pipeline. Dataset paths, image size, normalisation and class-balance rules sit in one place; the backbone, training loop and evaluation stay identical. You flip a dataset flag and the same script runs end to end. I would use a solid CNN for feature extraction, a light GAN module to grow the smaller class with realistic samples so the model does not lean toward the majority, then Grad-CAM overlays saved next to the originals so you can see whether the network is actually attending to tumour tissue. Each run would report accuracy, AUC and a confusion matrix. You get well-commented Python (PyTorch or TensorFlow, your pick), a clear config section, a short README for setup and reproduction, and a success check on at least one sample per modality with Grad-CAM landing on the right regions. If that sounds like what you need, I can get started right away.

User Avatar
Flag of INDIA

Python and software architecture are the strongest foundation for turning your four medical-imaging datasets into a reusable classification pipeline. I can develop a clean, configurable codebase that keeps the CNN training, validation, testing, metrics, and explainability logic consistent while allowing modality-specific preprocessing settings for MRI, ultrasound, mammogram, and histology data. My Python and AI-focused development background supports an end-to-end implementation with configurable dataset paths and parameters, accuracy, AUC, confusion-matrix outputs, GAN-based augmentation, and Grad-CAM visualisations saved alongside the source images. I will also provide well-commented code and a concise README covering environment setup and reproducible execution, with the architecture organised for maintainability and reuse. I am ready to build this CNN-GAN medical-image classifier and deliver a practical codebase that can be run across all four modalities.

User Avatar
Flag of GERMANY

Hello, "CNN-GAN Cancer Image Classifier" - the key is keeping one reusable training pipeline across four imaging modalities without letting preprocessing differences distort the evaluation. I'll structure the Python code around dataset configuration, consistent train/validation/test splits, modality-specific input preprocessing and a shared evaluation pipeline. I'll also account for class imbalance and data leakage: GAN-generated images must only augment the training set, never validation or test data. The deliverable will include reproducible configuration, accuracy, AUC, confusion matrices, saved predictions and Grad-CAM overlays, plus setup documentation. One important limitation: CNN, GAN, TensorFlow and PyTorch development are outside my confirmed core skills, so this is a weak technical match. I cannot honestly claim medical-image model development expertise. Would you be willing to accept a Python data-pipeline and evaluation implementation without requiring me to develop the CNN-GAN architecture itself? Looking forward to working with you. Beate

User Avatar
Flag of NIGERIA

A reusable end‑to‑end pipeline that combines a convolutional backbone, GAN‑based augmentation and Grad‑CAM visualization for any of the four medical imaging datasets is what you need. I will build a modular Python codebase using PyTorch that loads a dataset based on a simple configuration file, applies modality‑specific preprocessing, balances classes with a conditional GAN, trains a CNN classifier, evaluates accuracy, AUC and confusion matrix, and generates Grad‑CAM heat‑maps saved alongside the original images. The core architecture, training loop and evaluation logic remain unchanged while the config lets you switch datasets without touching the model code. All scripts will be documented, include command‑line arguments for paths and hyperparameters, and produce reproducible results with saved model checkpoints. Could you confirm the image file format and whether the labels are stored in separate CSV files? Let’s chat to lock in the exact scope and get started.

User Avatar
Flag of VIETNAM

Hello, Van Hieu here "MEDICAL IMAGE CLASSIFICATION" - reliable accuracy across four different imaging modalities Different image modalities can produce misleading accuracy if preprocessing and validation are not handled carefully. Grad-CAM also needs validation against actual tumour annotations to avoid visually convincing but incorrect explanations. How I would do it: 1. Build a configurable architecture with shared training and evaluation logic. 2. Adapt preprocessing to each modality without changing the core pipeline. 3. Integrate GAN augmentation while monitoring synthetic image quality. 4. Evaluate class performance using AUC, accuracy and confusion matrices. 5. Generate Grad-CAM overlays and save reproducible evaluation outputs. You get reusable, well-documented code that makes switching datasets straightforward and keeps results easy to inspect. My advantage is combining AI and data science to connect model performance with reliable evaluation. I bring 10 years of experience and 100+ completed projects. I suggest starting with one modality as a first milestone so you can review the code and results before proceeding. Do you already have tumour-region annotations available for checking the Grad-CAM results? I am looking forward to working with you.

User Avatar
Flag of PHILIPPINES

Hi there, The hard part is not just building a model to classify the datasets; it’s ensuring that the varying resolutions and appearances don’t compromise the overall accuracy. A unified architecture that allows for modality-specific preprocessing while maintaining consistent training and evaluation processes is crucial. I propose developing an end-to-end codebase in TensorFlow or PyTorch that incorporates a convolutional backbone for feature extraction, a GAN module for data augmentation, and Grad-CAM for visualization. The configuration section will allow for easy switching between datasets without altering the core logic. Could you clarify how you envision the structure of the README for environment setup? Thank you.

User Avatar
Flag of UKRAINE

Hi, I hope you’re doing well. This project fits well with my Python experience, and I can help build a reusable CNN-GAN pipeline for your medical image datasets. A clean TensorFlow or PyTorch implementation should let you switch between MRI, ultrasound, mammogram, and histology data while keeping the training and evaluation flow consistent. I would structure the code with a clear config layer for modality-specific preprocessing, then keep the backbone, GAN-based augmentation, and Grad-CAM logic modular so the same script can run across all four datasets. That makes it easier to maintain, compare results, and reproduce accuracy, AUC, confusion matrix, and heat-map outputs reliably. I’d start by reviewing the dataset structure and defining the shared pipeline, then implement the model, augmentation, and visualisation steps, test the key workflows on each modality, and provide a short README for setup and reproduction. I’d also be happy to help with fixes, updates, and ongoing maintenance as your needs grow. Best regards, Ruslan

User Avatar
Flag of UNITED ARAB EMIRATES

A PhD in Artificial Intelligence and nearly 20 years across research, software engineering, and technology leadership give me the depth to develop your reusable CNN-GAN medical-image classifier. My computer vision, deep learning, Python, PyTorch and machine learning expertise directly match the need for a unified pipeline across MRI, ultrasound, mammogram and histology data. I can build the configurable training and evaluation code around modality-specific preprocessing while keeping the core architecture, augmentation, validation and testing logic consistent. The implementation will include benign/malignant classification, GAN-based augmentation, accuracy, AUC and confusion-matrix reporting, along with Grad-CAM visualisations saved beside the corresponding images. My software architecture background also supports a clean configuration section, reproducible scripts or notebook, comments and README documentation. My experience leading end-to-end AI product development enables me to deliver a practical, maintainable codebase rather than an isolated model experiment. I would be pleased to apply my expertise to this project and deliver the complete working pipeline.

User Avatar
Flag of VIETNAM

Hello, My strongest match for this project is hands-on Python engineering for production AI and document-processing pipelines. I build reliable, reusable codebases rather than one-off prototypes, with experience handling complex, messy inputs and making workflows reproducible across different data sources. For your cancer-image classifier, I can apply that engineering discipline to a configurable Python pipeline covering dataset selection, modality-specific preprocessing, CNN training, GAN-based augmentation, validation and testing. The implementation can keep the core model and evaluation logic reusable while exposing dataset paths and preprocessing settings in a clear configuration section. I can also structure the outputs around accuracy, AUC, confusion matrices and saved Grad-CAM overlays, with commented code and setup documentation. My background includes Python, AI systems, OCR and HIPAA-conscious architecture, which is relevant to building an organized medical-imaging workflow with careful handling of sensitive data. I would be glad to build and document this CNN-GAN explainability pipeline so it runs consistently across your four modalities. Thanks!

User Avatar
Flag of SERBIA

Hi, I can build a reusable Python CNN-GAN cancer image classification pipeline for MRI, ultrasound, mammogram, and histology datasets, with modality-specific configuration and shared training and evaluation logic. I will structure the code around a clear configuration section for paths, image size, augmentation, and training parameters. The pipeline will use a CNN backbone, GAN-based class balancing and image generation, and Grad-CAM overlays saved beside the source images. It will report accuracy, AUC, confusion matrices, and train/validation/test results, with validation on a sample from each modality. A commented notebook or script and reproducible README will be included. Grad-CAM can show discriminative regions but does not by itself prove clinical localization, so I will keep preprocessing, splits, and evaluation reproducible and clearly documented. Thank you

内容来自 Freelancer News,版权归原作者所有。

阅读原文 ↗

把喜欢的内容,留在你的新标签页。

打开我的订阅