Transformer-based end-to-end classification of variable-length volumetric data

TL;DR

This paper introduces an efficient Transformer-based method for classifying variable-length 3D medical volumes, improving robustness and accuracy by randomizing input resolution during training.

Contribution

It proposes a novel training strategy that enhances Transformer models' ability to handle variable-length volumetric data without losing relevant information.

Findings

Achieved 21.96% improvement in balanced accuracy on retinal OCT classification.

Randomizing input resolution during training improves volume representation.

Model is more robust to variable volume length and different computational budgets.

Abstract

The automatic classification of 3D medical data is memory-intensive. Also, variations in the number of slices between samples is common. Na\"ive solutions such as subsampling can solve these problems, but at the cost of potentially eliminating relevant diagnosis information. Transformers have shown promising performance for sequential data analysis. However, their application for long sequences is data, computationally, and memory demanding. In this paper, we propose an end-to-end Transformer-based framework that allows to classify volumetric data of variable length in an efficient fashion. Particularly, by randomizing the input volume-wise resolution(#slices) during training, we enhance the capacity of the learnable positional embedding assigned to each volume slice. Consequently, the accumulated positional information in each positional embedding can be generalized to the neighbouring slices, even for high-resolution volumes at the test time. By doing so, the model will be more robust to variable volume length and amenable to different computational budgets. We evaluated the proposed approach in retinal OCT volume classification and achieved 21.96% average improvement in balanced accuracy on a 9-class diagnostic task, compared to state-of-the-art video transformers. Our findings show that varying the volume-wise resolution of the input during training results in more informative volume representation as compared to training with fixed number of slices per volume.