-
FCN-Transformer Feature Fusion for Polyp Segmentation
Authors:
Edward Sanderson,
Bogdan J. Matuszewski
Abstract:
Colonoscopy is widely recognised as the gold standard procedure for the early detection of colorectal cancer (CRC). Segmentation is valuable for two significant clinical applications, namely lesion detection and classification, providing means to improve accuracy and robustness. The manual segmentation of polyps in colonoscopy images is time-consuming. As a result, the use of deep learning (DL) fo…
▽ More
Colonoscopy is widely recognised as the gold standard procedure for the early detection of colorectal cancer (CRC). Segmentation is valuable for two significant clinical applications, namely lesion detection and classification, providing means to improve accuracy and robustness. The manual segmentation of polyps in colonoscopy images is time-consuming. As a result, the use of deep learning (DL) for automation of polyp segmentation has become important. However, DL-based solutions can be vulnerable to overfitting and the resulting inability to generalise to images captured by different colonoscopes. Recent transformer-based architectures for semantic segmentation both achieve higher performance and generalise better than alternatives, however typically predict a segmentation map of $\frac{h}{4}\times\frac{w}{4}$ spatial dimensions for a $h\times w$ input image. To this end, we propose a new architecture for full-size segmentation which leverages the strengths of a transformer in extracting the most important features for segmentation in a primary branch, while compensating for its limitations in full-size prediction with a secondary fully convolutional branch. The resulting features from both branches are then fused for final prediction of a $h\times w$ segmentation map. We demonstrate our method's state-of-the-art performance with respect to the mDice, mIoU, mPrecision, and mRecall metrics, on both the Kvasir-SEG and CVC-ClinicDB dataset benchmarks. Additionally, we train the model on each of these datasets and evaluate on the other to demonstrate its superior generalisation performance.
△ Less
Submitted 17 August, 2022;
originally announced August 2022.
-
mREAL-GAN: Generating Multiple Residential Electrical Appliance Load Profiles with Inter-Dependencies using a Generative Adversarial Network
Authors:
Edward Sanderson,
Aikaterini Fragaki,
Jules Simo,
Bogdan J. Matuszewski
Abstract:
In this paper, we introduce mREAL-GAN, a generative adversarial network (GAN) for the parallel generation of multiple residential electrical appliance load (mREAL) profiles. mREAL-GAN is intended for use in community-scale low-voltage network analysis, and represents a departure from previous methods for this purpose, which break the generation of appliance load profiles into several steps and lar…
▽ More
In this paper, we introduce mREAL-GAN, a generative adversarial network (GAN) for the parallel generation of multiple residential electrical appliance load (mREAL) profiles. mREAL-GAN is intended for use in community-scale low-voltage network analysis, and represents a departure from previous methods for this purpose, which break the generation of appliance load profiles into several steps and largely model each appliance independently. Instead, mREAL-GAN models appliance load profiles in an end-to-end manner, and generates multiple appliance load profiles in parallel in a way that captures inter-dependencies. We show that mREAL-GAN generates load profiles for individual appliance-types with greater fidelity than a popular example of previous methods, and demonstrate its ability to capture inter-dependencies between appliances.
△ Less
Submitted 8 December, 2021;
originally announced December 2021.
-
Inter-Patient ECG Classification with Convolutional and Recurrent Neural Networks
Authors:
Li Guo,
Gavin Sim,
Bogdan Matuszewski
Abstract:
The recent advances in ECG sensor devices provide opportunities for user self-managed auto-diagnosis and monitoring services over the internet. This imposes the requirements for generic ECG classification methods that are inter-patient and device independent. In this paper, we present our work on using the densely connected convolutional neural network (DenseNet) and gated recurrent unit network (…
▽ More
The recent advances in ECG sensor devices provide opportunities for user self-managed auto-diagnosis and monitoring services over the internet. This imposes the requirements for generic ECG classification methods that are inter-patient and device independent. In this paper, we present our work on using the densely connected convolutional neural network (DenseNet) and gated recurrent unit network (GRU) for addressing the inter-patient ECG classification problem. A deep learning model architecture is proposed and is evaluated using the MIT-BIH Arrhythmia and Supraventricular Databases. The results obtained show that without applying any complicated data pre-processing or feature engineering methods, both of our models have considerably outperformed the state-of-the-art performance for supraventricular (SVEB) and ventricular (VEB) arrhythmia classifications on the unseen testing dataset (with the F1 score improved from 51.08 to 61.25 for SVEB detection and from 88.59 to 89.75 for VEB detection respectively). As no patient-specific or device-specific information is used at the training stage in this work, it can be considered as a more generic approach for dealing with scenarios in which varieties of ECG signals are collected from different patients using different types of sensor devices.
△ Less
Submitted 27 September, 2018;
originally announced October 2018.
