ICM-BTD: improved classification model for brain tumor diagnosis using discrete wavelet transform-based feature extraction and SVM classifier

SG-Transunet: A segmentation-guided Transformer U-Net model for KRAS gene mutation status identification in colorectal cancer

Accurately identifying the Kirsten rat sarcoma virus (KRAS) gene mutation status in colorectal cancer (CRC) patients can assist doctors in deciding whether to use specific targeted drugs for treatment. Although deep learning methods are popular, they are often affected by redundant features from non-lesion areas. Moreover, existing methods commonly extract spatial features from imaging data, which neglect important frequency domain features and may degrade the performance of KRAS gene mutation status identification. To address this deficiency, we propose a segmentation-guided Transformer U-Net (SG-Transunet) model for KRAS gene mutation status identification in CRC. Integrating the strength of convolutional neural networks (CNNs) and Transformers, SG-Transunet offers a unique approach for both lesion segmentation and KRAS mutation status identification. Specifically, for precise lesion localization, we employ an encoder-decoder to obtain segmentation results and guide the KRAS gene mutation status identification task. Subsequently, a frequency domain supplement block is designed to capture frequency domain features, integrating it with high-level spatial features extracted in the encoding path to derive advanced spatial-frequency domain features. Furthermore, we introduce a pre-trained Xception block to mitigate the risk of overfitting associated with small-scale datasets. Following this, an aggregate attention module is devised to consolidate spatial-frequency domain features with global information extracted by the Transformer at shallow and deep levels, thereby enhancing feature discriminability. Finally, we propose a mutual-constrained loss function that simultaneously constrains the segmentation mask acquisition and gene status identification process. Experimental results demonstrate the superior performance of SG-Transunet over state-of-the-art methods in discriminating KRAS gene mutation status.

A Novel Lightweight CNN Architecture for the Diagnosis of Brain Tumors Using MR Images

Over the last few years, brain tumor-related clinical cases have increased substantially, particularly in adults, due to environmental and genetic factors. If they are unidentified in the early stages, there is a risk of severe medical complications, including death. So, early diagnosis of brain tumors plays a vital role in treatment planning and improving a patient's condition. There are different forms, properties, and treatments of brain tumors. Among them, manual identification and classification of brain tumors are complex, time-demanding, and sensitive to error. Based on these observations, we developed an automated methodology for detecting and classifying brain tumors using the magnetic resonance (MR) imaging modality. The proposed work includes three phases: pre-processing, classification, and segmentation. In the pre-processing, we started with the skull-stripping process through morphological and thresholding operations to eliminate non-brain matters such as skin, muscle, fat, and eyeballs. Then we employed image data augmentation to improve the model accuracy by minimizing the overfitting. Later in the classification phase, we developed a novel lightweight convolutional neural network (lightweight CNN) model to extract features from skull-free augmented brain MR images and then classify them as normal and abnormal. Finally, we obtained infected tumor regions from the brain MR images in the segmentation phase using a fast-linking modified spiking cortical model (FL-MSCM). Based on this sequence of operations, our framework achieved 99.58% classification accuracy and 95.7% of dice similarity coefficient (DSC). The experimental results illustrate the efficiency of the proposed framework and its appreciable performance compared to the existing techniques.

Discrete-MultiResUNet: Segmentation and feature extraction model for knee MR images

Deep learning has shown outstanding efficiency in medical image segmentation. Segmentation of knee tissues is an important task for early diagnosis of rheumatoid arthritis (RA) with selecting variant features. Automated segmentation and feature extraction of knee tissues are desirable for faster and reliable analysis of large datasets and further diagnosis. In this paper a novel architecture called as Discrete-MultiResUNet, which is a combination of discrete wavelet transform (DWT) with MultiResUNet architecture is applied for feature extraction and segmentation, respectively. This hybrid architecture captures more prominent features from the knee magnetic resonance image efficiently with segmenting vital knee tissues. The hybrid model is evaluated on the knee MR dataset demonstrating outperforming performance compared with baseline models. The model achieves excellent segmentation performance accuracy of 96.77% with a dice coefficient of 98%.

Spatial-Frequency dual-branch attention model for determining KRAS mutation status in colorectal cancer with T2-weighted MRI

BACKGROUND AND OBJECTIVE

Identifying the KRAS mutation status accurately in medical images is very important for the diagnosis and treatment of colorectal cancer. Despite the substantial progress achieved by existing methods, it remains challenging due to limited annotated dataset, large intra-class variances, and a high degree of inter-class similarities.

METHODS

To tackle these challenges, we propose a spatial-frequency dual-branch attention model (SF-DBAM) to determine the KRAS mutation status of colorectal cancer patients using a limited T2-weighted MRI dataset. The dataset contains 169 wild-type patients (2151 images) and 137 mutation-type patients (1666 images). The first branch utilizes part of the pre-trained Xception model to capture spatial-domain information and alleviate the small-scale dataset problem. The second branch builds frequency-domain information into cube columns using block-based discrete cosine transform and channel rearrangement. Then the cube columns are fed into convolutional long short-term memory (convLSTM) to explore the effective information between the reconstructed frequency-domain channels. Also, we design a channel enhanced attention module (CEAM) at the end of each branch to make them focus on the lesion areas. Finally, we concatenate the two branches and output the classified results through fully connected layers.

RESULTS

The proposed method achieves 88.03% overall accuracy with AUC of 94.27% and specificity of 90.75% in 10-fold cross-validation, which is better than the current non-invasive methods for determining KRAS mutation status.

CONCLUSIONS

We believe that the proposed method can assist physicians to diagnose the KRAS mutation status in patients with colorectal cancer, and other medical problems can benefit from the spatial and frequency domains information.

Detection of changes in the circadian rhythm of cattle in relation to disease, stress, and reproductive events

Disease and stress can disrupt the circadian rhythm of activity in animals. Sensor technologies can automatically detect variations in daily activity, but it remains difficult to detect exactly when the circadian rhythm disruption starts. Here we report a mathematical Fourier-Based Approximation with Thresholding (FBAT) method designed to detect changes in the circadian activity rhythm of cows whatever the cause of change (typically disease, stress, oestrus). We used data from an indoor positioning system that provides the time per hour spent by each cow resting, in alleys, or eating. We calculated the hourly activity level of each cow by attributing a weight to each activity. We considered 36-h time series and used Fourier transform to model the variations in activity during the first and last 24 h of these 36-h series. We then compared the Euclidian distance between the two models against a given threshold above which we considered that rhythm had changed. We tested the method on four datasets (giving a cumulative total of ~120000 cow*days) that included disease episodes (acidosis, lameness, mastitis or other infectious diseases), reproductive events (oestrus or calving) and external stimuli that can stress animals (e.g. relocation). The method obtained over 80% recall of normal days and detected 95% of abnormal rhythms due to health or reproductive events. FBAT could be implemented in precision livestock farming system monitoring tools to alert caretakers to individual animals needing specific care. The FBAT method also has the potential to detect anomalies in humans to guide healthcare intervention or in wild animals to detect disturbances. We anticipate that chronobiological studies could apply FBAT to help relate circadian rhythm anomalies to specific events.