Multimodal Microscopic Spectrum-Image Analysis and Intelligent Fusion Diagnosis of Breast Cancer

Qingxia Wu; Bainan Li; Ziyang Hui; Zihan Wang; Yunhong Li; Linwei Shang; Jianhua Yin

doi:10.3788/CJL240724

Chinese Journal of Lasers, Volume. 51, Issue 15, 1507205(2024)

Multimodal Microscopic Spectrum-Image Analysis and Intelligent Fusion Diagnosis of Breast Cancer

Qingxia Wu, Bainan Li, Ziyang Hui, Zihan Wang, Yunhong Li, Linwei Shang, and Jianhua Yin^*

Department of Biomedical Engineering, College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, Jiangsu , China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(10)

Fig. 1. Schematic diagram of system light path

Download full size

Fig. 2. Spectrum‒image fusion neural network

Download full size

Fig. 3. Bright field, blue fluorescence and fusion images of normal and cancerous breast tissues, respectively

Download full size

Fig. 4. Bright field, green fluorescence and fusion images of normal and cancerous breast tissues, respectively

Download full size

Fig. 5. Fluorescence spectra of breast tissues. (a) Average fluorescence spectra of breast tissue with purple light excitation; (b) normalized fluorescence spectra of breast tissue with blue light excitation

Download full size

Fig. 6. Fluorescence spectra and Gaussian-fitted peaks of breast tissues. (a) Normal breast tissue; (b) cancerous breast tissue

Download full size

Fig. 7. Results from fusion network model. (a) Loss function and accuracy curves; (b) ROC curve; (c) confusion matrices

Download full size

Table 1. Characteristic parameter information of Gaussian fitting peaks
View table
Table 1. Characteristic parameter information of Gaussian fitting peaks
Classification A₁ A₂ A₃ λ₁ /nm λ₂ /nm λ₃ /nm ω₁ ω₂ ω₃ R²
Cancerous 30.89 89.66 12.06 470 520 635 65.22 118.56 99.80 0.993
Normal 40.97 72.02 7.73 470 520 635 68.64 113.21 96.19 0.993

Table 2. Peak area ratios of normal and cancerous breast tissues
View table
Table 2. Peak area ratios of normal and cancerous breast tissues
Classification A₅₂₀/A₄₇₀ A₆₃₅/A₄₇₀
Cancerous 2.9025 0.3903
Normal 1.7578 0.1886

Table 3. Comparison of the diagnostic results of the spectrum‒image fusion dataset with those of 5 types of individual images and spectral data

View table

Table 3. Comparison of the diagnostic results of the spectrum‒image fusion dataset with those of 5 types of individual images and spectral data

Type of data	Model	Accuracy /%
Type of data	Model	Training set	Validation set	Test set
Bright-field image	2D-CNN	95.93±5.20	93.47±3.94	80.22±7.79
Fluorescence image with purple light excitation	2D-CNN	94.60±5.14	91.83±4.74	79.92±8.25
Fluorescence image with blue light excitation	2D-CNN	91.27±5.80	90.90±9.30	81.43±11.87
Fluorescence spectrum with purple light excitation	1D-CNN	88.23±8.63	88.32±7.01	72.41±13.75
Fluorescence spectrum with blue light excitation	1D-CNN	93.51±5.00	90.19±9.39	76.94±7.71
Spectrum‒image fusion	Spectrum‒image fusion neural network	92.98±4.44	93.49±4.63	86.38±11.62

Tools

Get Citation

Copy Citation Text

Qingxia Wu, Bainan Li, Ziyang Hui, Zihan Wang, Yunhong Li, Linwei Shang, Jianhua Yin. Multimodal Microscopic Spectrum-Image Analysis and Intelligent Fusion Diagnosis of Breast Cancer[J]. Chinese Journal of Lasers, 2024, 51(15): 1507205

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites