# Scopeformer: n-CNN-ViT Hybrid Model for Intracranial Hemorrhage Classification

arXiv: https://arxiv.org/abs/2107.04575 · HTML: https://mlsanity.com/paper/2107.04575/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 1 of its 3 references that are in the map.
- First posted on arXiv 7 Jul 2021.

## Builds on (most cited first, 1 of 1)

- [Accurate and Efficient Intracranial Hemorrhage Detection and Subtype Classification in 3D CT Scans with Convolutional and Long Short-Term Memory Neural Networks](https://mlsanity.com/paper/2008.00302/) (2020)

## Built on by (most cited first, 7 of 7)

- [Transformers in Medical Image Analysis: A Review](https://mlsanity.com/paper/2202.12165/) (2022)
- [A survey on attention mechanisms for medical applications: are we moving towards better algorithms?](https://mlsanity.com/paper/2204.12406/) (2022)
- [Vision Transformers in Medical Imaging: A Review](https://mlsanity.com/paper/2211.10043/) (2022)
- [MoViT: Memorizing Vision Transformers for Medical Image Analysis](https://mlsanity.com/paper/2303.15553/) (2023)
- [Efficient Scopeformer: Towards Scalable and Rich Feature Extraction for Intracranial Hemorrhage Detection](https://mlsanity.com/paper/2302.00220/) (2023)
- [A Novel Vision Transformer with Residual in Self-attention for Biomedical Image Classification](https://mlsanity.com/paper/2306.01594/) (2023)
- [Multi-Plane Vision Transformer for Hemorrhage Classification Using Axial and Sagittal MRI Data](https://mlsanity.com/paper/2505.07349/) (2025)

## Abstract (arXiv)

We propose a feature generator backbone composed of an ensemble of convolutional neuralnetworks (CNNs) to improve the recently emerging Vision Transformer (ViT) models. We tackled the RSNA intracranial hemorrhage classification problem, i.e., identifying various hemorrhage types from computed tomography (CT) slices. We show that by gradually stacking several feature maps extracted using multiple Xception CNNs, we can develop a feature-rich input for the ViT model. Our approach allowed the ViT model to pay attention to relevant features at multiple levels. Moreover, pretraining the n CNNs using various paradigms leads to a diverse feature set and further improves the performance of the proposed n-CNN-ViT. We achieved a test accuracy of 98.04% with a weighted logarithmic loss value of 0.0708. The proposed architecture is modular and scalable in both the number of CNNs used for feature extraction and the size of the ViT.