Ivan Drokin

Ivan Drokin

The emergence of Kolmogorov-Arnold Networks (KANs) has sparked significant interest and debate within the scientific community. This paper explores the application of KANs in the domain of computer vision (CV). We examine the convolutional version of KANs, considering various nonlinearity options beyond splines, such as Wavelet transforms and a range of polynomials. We propose a parameter-efficient design for Kolmogorov-Arnold convolutional layers and a parameter-efficient finetuning algorithm for pre-trained KAN models, as well as KAN convolutional versions of self-attention and focal modulation layers. We provide empirical evaluations conducted on MNIST, CIFAR10, CIFAR100, Tiny ImageNet, ImageNet1k, and HAM10000 datasets for image classification tasks. Additionally, we explore segmentation tasks, proposing U-Net-like architectures with KAN convolutions, and achieving state-of-the-art results on BUSI, GlaS, and CVC datasets. We summarized all of our findings in a preliminary design guide of KAN convolutional models for computer vision tasks. Furthermore, we investigate regularization techniques for KANs. All experimental code and implementations of convolutional layers and models, pre-trained on ImageNet1k weights are available on GitHub via this https://github.com/IvanDrokin/torch-conv-kan

Ivan Drokin, Elena Ericheva

This paper proposes novel end-to-end framework for detecting suspicious pulmonary nodules in chest CT scans. The method core idea is a new nodule segmentation architecture with a model-based feature projection block on three-dimensional convolutions. This block acts as a preliminary feature extractor for a two-dimensional U-Net-like convolutional network. Using the proposed approach along with an axial, coronal, and sagittal projection analysis makes it possible to abandon the widely used false positives reduction step. The proposed method achieves SOTA on LUNA2016 with 0.959 average sensitivity, and 0.936 sensitivity if the false-positive level per scan is 0.25. The paper describes the proposed approach and represents the experimental results on LUNA2016 as well as ablation studies.

Ivan Drokin, Elena Ericheva

This paper focuses on a novel approach for false-positive reduction (FPR) of nodule candidates in Computer-aided detection (CADe) systems following the suspicious lesions detection stage. Contrary to typical decisions in medical image analysis, the proposed approach considers input data not as a 2D or 3D image, but rather as a point cloud, and uses deep learning models for point clouds. We discovered that point cloud models require less memory and are faster both in training and inference compared to traditional CNN 3D, they achieve better performance and do not impose restrictions on the size of the input image, i.e. no restrictions on the size of the nodule candidate. We propose an algorithm for transforming 3D CT scan data to point cloud. In some cases, the volume of the nodule candidate can be much smaller than the surrounding context, for example, in the case of subpleural localization of the nodule. Therefore, we developed an algorithm for sampling points from a point cloud constructed from a 3D image of the candidate region. The algorithm is able to guarantee the capture of both context and candidate information as part of the point cloud of the nodule candidate. We designed and set up an experiment in creating a dataset from an open LIDC-IDRI database for a feature of the FPR task, and is herein described in detail. Data augmentation was applied both to avoid overfitting and as an upsampling method. Experiments were conducted with PointNet, PointNet++, and DGCNN. We show that the proposed approach outperforms baseline CNN 3D models and resulted in 85.98 FROC versus 77.26 FROC for baseline models. We compare our algorithm with published SOTA and demonstrate that even without significant modifications it works at the appropriate performance level on LUNA2016 and shows SOTA on LIDC-IDRI.

Tatiana Malygina, Elena Ericheva, Ivan Drokin

We propose a novel method to improve deep learning model performance on highly-imbalanced tasks. The proposed method is based on CycleGAN to achieve balanced dataset. We show that data augmentation with GAN helps to improve accuracy of pneumonia binary classification task even if the generative network was trained on the same training dataset.