Cross Pseudo Supervision Algorithm for Identifying Neuroblastoma Differentiation Type in Whole Slide Pathology Image

Zhenzhen Wan; Yuwei Liu; Ning Shi; Haocheng Li; Fang Liu

doi:10.3788/CJL240489

Chinese Journal of Lasers, Volume. 51, Issue 15, 1507107(2024)

Cross Pseudo Supervision Algorithm for Identifying Neuroblastoma Differentiation Type in Whole Slide Pathology Image

Zhenzhen Wan¹, Yuwei Liu¹, Ning Shi^1、*, Haocheng Li¹, and Fang Liu^2、**

¹Key Laboratory of Digital Medical Engineering of Hebei Province, College of Electronic and Information Engineering, Hebei University, Baoding 071002, Hebei , China

²Department of Pathology, Baoding Hospital of Beijing Children’s Hospital, Capital Medical University, Baoding , Hebei 071002, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(13)

Fig. 1. Schematic diagram of CPS model

Download full size

Fig. 2. Schematic diagram of CSA module

Download full size

Fig. 3. Schematic diagram of channel attention mechanism

Download full size

Fig. 4. Schematic diagram of spatial attention mechanism

Download full size

Fig. 5. Diagram of CSA-U-Net network structure

Download full size

Fig. 6. Flow chart of K-means cell count

Download full size

Fig. 7. Schematic diagram of different cell types

Download full size

Fig. 8. Comparison results of segmentation and classification of NB images by different methods

Download full size

Fig. 9. Cell counting results

Download full size

Table 1. Comparison of experimental results of different methods on NB dataset

View table

Table 1. Comparison of experimental results of different methods on NB dataset

Algorithm	F1-score /%		Acc /%
Algorithm	Poorly differentiated	Differentiated	Acc /%
U-Net	77.99	59.59	96.71
DeepLabv3+	64.59	24.02	95.60
PSPNet	31.24	42.29	94.45
HrNet	75.05	47.89	96.76
SA-U-Net	78.07	59.71	96.66
HoVer-Net	69.28	50.81	94.77
MEDIAR	68.59	51.96	96.38
CSA-U-Net	79.05	62.21	96.78

Table 2. Comparison of experimental results of labeled and unlabeled data in different proportions
View table
Table 2. Comparison of experimental results of labeled and unlabeled data in different proportions
Proportion Labeled Unlabeled F1-score /%
Acc /
%
Poorly differentiated Differentiated
1∶1 1200 1200 79.21 62.53 97.64
1∶2 1200 2400 80.52 64.38 97.84
1∶3 1200 3600 80.99 65.40 97.99
1∶4 1200 4800 81.02 65.48 98.02

Table 3. Comparison of experimental results of different semi-supervised methods
View table
Table 3. Comparison of experimental results of different semi-supervised methods
Method F1-score /%
Acc /
%
Poorly differentiated Differentiated
Self-training 76.63 58.44 96.04
Mean teacher 78.72 60.88 96.71
GCT 73.58 57.18 96.67
CPS 80.99 65.40 97.99

Table 4. Counting accuracy in poorly differentiated and differentiated cells
View table
Table 4. Counting accuracy in poorly differentiated and differentiated cells
No. Poorly differentiated /% Differentiated /%
1 100.00 83.33
2 92.85 100.00
3 92.59 100.00
4 88.10 80.00
5 94.29 92.00
6 93.33 82.35
7 88.88 75.00
8 95.88 94.12
9 98.57 100.00
10 95.54 92.11

Tools

Get Citation

Copy Citation Text

Zhenzhen Wan, Yuwei Liu, Ning Shi, Haocheng Li, Fang Liu. Cross Pseudo Supervision Algorithm for Identifying Neuroblastoma Differentiation Type in Whole Slide Pathology Image[J]. Chinese Journal of Lasers, 2024, 51(15): 1507107

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Biomedical Optical Imaging

Received: Jan. 12, 2024

Accepted: Mar. 11, 2024

Published Online: Jul. 23, 2024

The Author Email: Ning Shi (shiningzhongguo@126.com), Fang Liu (asasfang@163.com)

DOI:10.3788/CJL240489

CSTR:32183.14.CJL240489

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Comparison of experimental results of different methods on NB dataset

Table 1. Comparison of experimental results of different methods on NB dataset

Table 2. Comparison of experimental results of labeled and unlabeled data in different proportions

Table 2. Comparison of experimental results of labeled and unlabeled data in different proportions

Table 3. Comparison of experimental results of different semi-supervised methods

Table 3. Comparison of experimental results of different semi-supervised methods

Table 4. Counting accuracy in poorly differentiated and differentiated cells

Table 4. Counting accuracy in poorly differentiated and differentiated cells