PPubMed17 Mar 2022Cited 6×

Diagnosis System of Microscopic Hyperspectral Image of Hepatobiliary Tumors Based on Convolutional Neural Network

Shixin Huang, Jiawei Luo, Kexue Pu, Min Wu

Abstract

Hepatobiliary tumor is one of the common tumors and cancers in medicine, which seriously affects people's lives, so how to accurately diagnose it is a very serious problem. This article mainly studies a diagnostic method of microscopic images of liver and gallbladder tumors. Under this research direction, this article proposes to use convolutional neural network to learn and use hyperspectral images to diagnose it. It is found that the addition of the convolutional neural network can greatly improve the actual map classification and the accuracy of the map, and effectively improve the success rate of the treatment. At the same time, the article designs related experiments to compare its feature extraction performance and classification situation. The experimental results in this article show that the improved diagnostic method based on convolutional neural network has an accuracy rate of 85%-90%, which is as high as 6%-8% compared with the traditional accuracy rate, and thus it effectively improves the clinical problem of hepatobiliary tumor treatment.

A figure from Diagnosis System of Microscopic Hyperspectral Image of Hepatobiliary Tumors Based on Convolutional Neural Network — fig. from the paper

§ The Valyu brief

The paper, in five minutes

Read from the full text via Valyu Contents

A CNN-based system using microscopic hyperspectral images diagnoses hepatobiliary tumors with 85‑90% accuracy, 6‑8% better than traditional methods.

What’s new

i.Proposed a convolutional neural network (SDCNN) to diagnose hepatobiliary tumors from microscopic hyperspectral images.
ii.Introduced adaptive convolution kernel generation using MCFSFDP clustering for improved feature extraction.
iii.Demonstrated a 6‑8% accuracy improvement over traditional diagnostic methods on 100 patient cases.
iv.Conducted extensive experiments on public hyperspectral datasets (Indian Pines, Pavia University) showing superior OA, AA, and Kappa scores.
v.Integrated spatial and spectral information via a 3D CNN architecture with batch normalization.

How they did it

The workflow preprocesses microscopic hyperspectral images, clusters sub‑image blocks with MCFSFDP to create adaptive convolution kernels, and feeds these kernels into a three‑dimensional CNN (SDCNN) that learns joint spatial‑spectral features. The network is trained and validated on public hyperspectral datasets and on 100 hepatobiliary tumor patient samples, and its diagnostic performance is compared to traditional methods.

What they found

Diagnostic accuracy on 100 hepatobiliary tumor cases: 85‑90% (CNN) vs 79‑82% (traditional), a 6‑8% improvement.
Indian Pines dataset: OA 97.71%, AA 96.87%, Kappa 97.39% – outperforming SVM, 3DCNN, BASE‑3DCNN, and SSRN by 9%, 6.1%, 1.5%, and 0.5% respectively.
Pavia University dataset: OA 99.11%, AA 98.32%, Kappa 98.82% – surpassing SVM, 3DCNN, BASE‑3DCNN, and SSRN by 4.7%, 3.3%, 2.1%, and 0.5% respectively.

Caveats

✕Experiments were conducted under ideal laboratory conditions, limiting assessment of real‑world feasibility.
✕The clinical evaluation involved only 100 patient cases, which may not capture broader population variability.
✕High dimensionality of hyperspectral data can lead to increased computational requirements.
✕The study focuses on microscopic hyperspectral images; applicability to other imaging modalities remains untested.

§ The Valyu brief

Reading the full paper and taking notes. This takes a few seconds…

§ Ask this paper

Ask a question about this paper

Valyu reads the full text and answers from what the paper actually says.

§ Related work in other archives

Similar papers in all four archives

The same idea searched in arXiv, PubMed, bioRxiv and chemRxiv at once, so a method can lead you to where it is being applied.

AarXiv4 found

PPubMed4 found

BbioRxiv4 found

CchemRxiv2 found

Searching the other archives…