CONSULT: Contrastive Self-Supervised Learning for Few-shot Tumor Detection

Sin Chee Chin,Xuan Zhang,Lee Yeong Khang,Wenming Yang

2024-10-15

Abstract:Artificial intelligence aids in brain tumor detection via MRI scans, enhancing the accuracy and reducing the workload of medical professionals. However, in scenarios with extremely limited medical images, traditional deep learning approaches tend to fail due to the absence of anomalous images. Anomaly detection also suffers from ineffective feature extraction due to vague training process. Our work introduces a novel two-stage anomaly detection algorithm called CONSULT (CONtrastive Self-sUpervised Learning for few-shot Tumor detection). The first stage of CONSULT fine-tunes a pre-trained feature extractor specifically for MRI brain images, using a synthetic data generation pipeline to create tumor-like data. This process overcomes the lack of anomaly samples and enables the integration of attention mechanisms to focus on anomalous image segments. The first stage is to overcome the shortcomings of current anomaly detection in extracting features in high-variation data by incorporating Context-Aware Contrastive Learning and Self-supervised Feature Adversarial Learning. The second stage of CONSULT uses PatchCore for conventional feature extraction via the fine-tuned weights from the first stage. To summarize, we propose a self-supervised training scheme for anomaly detection, enhancing model performance and data reliability. Furthermore, our proposed contrastive loss, Tritanh Loss, stabilizes learning by offering a unique solution all while enhancing gradient flow. Finally, CONSULT achieves superior performance in few-shot brain tumor detection, demonstrating significant improvements over PatchCore by 9.4%, 12.9%, 10.2%, and 6.0% for 2, 4, 6, and 8 shots, respectively, while training exclusively on healthy images.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

### What problems does this paper attempt to solve?

This paper aims to solve the problem that traditional deep - learning methods cannot effectively detect tumors when there are very few medical images, especially brain MRI images. Specifically:

1. **Data scarcity**: In the field of medical imaging, it is very difficult to obtain enough abnormal samples (such as images with tumors), which leads to the problem of data scarcity. Traditional deep - learning models require a large amount of labeled data for training, but in practical applications, especially in brain MRI images, it is almost impossible to obtain enough abnormal samples.

2. **Inadequate feature extraction**: Existing anomaly detection methods are not effective in dealing with highly variable data, especially when there is no clear objective function, it is difficult to fine - tune the downstream anomaly detection algorithms, resulting in ineffective feature extraction.

3. **Cost and efficiency issues**: Interpreting MRI images requires highly specialized medical personnel, which not only increases the cost but also makes the whole process time - consuming and dependent on the experience of experts.

To solve these problems, the author proposes a two - stage anomaly detection algorithm named **CONSULT** (CONtrastive Self - sUpervised Learning for few - shot Tumor detection). The main contributions of this algorithm are as follows:

- **Self - supervised training scheme**: Fine - tune the feature extractor through self - supervised learning to adapt it to the field of MRI images, thereby reducing the need for a large number of training samples.

- **Contrastive loss function**: A new contrastive loss function - Tritanh Loss is introduced, which provides a more stable gradient flow and has a unique solution, thereby improving the stability of training.

- **Context - aware attention mechanism**: A lightweight attention module is added to the feature extractor, enabling the model to focus on abnormal regions, thereby enhancing the robustness and accuracy of feature extraction.

- **Performance improvement**: The experimental results show that CONSULT performs significantly better than PatchCore in few - shot brain tumor detection, with improvements of 9.4%, 12.9%, 10.2% and 6.0% in the cases of 2, 4, 6, and 8 shots respectively.

Through these innovations, CONSULT can achieve efficient and accurate brain tumor detection even with only healthy images, thus solving the challenges of existing methods in data scarcity and feature extraction.