Appliciation Of Multispectral Imaging In Histopathological Image Analysis Of Breast Cancer And Colorectal Cancer

Part 1.Application of multispectral imaging in quantitative immunohistochemistry study of breast cancer:A comparative studyBackground Multispectral imaging(MSI)based on imaging and spectroscopy,as relatively novel to the field of histopathology,has been used in biomedical multidisciplinary researches.This study aimed at analyzing and comparimg the utility of conventional red-green-blue(RGB)versus bright-field multispectral(MS)images for immunohistochemistry(IHC)staining of invasive breast cancer(BC)to explore the advantages of MSI in the clinical pathological diagnosis.Methods The corresponding RGB and MS images of conventional IHC-stained membranous marker HER2,cytoplasmic marker CK5/6,and nuclear marker ER of BC were acquired by traditional CCD and MSI systems.The total signal optical density(OD)values of the positive regions for each biomarker were quantitatively analyzed and compared using CRi Nuance image analysis software.Results The bright-field MS images acquired of IHC staining for three biomarkers have higher resolution,stronger contrast and more accurate segmentation than RGB images,making it easy for positive signal separation and quantification.The total signal OD values for each biomarker were higher in MS images than in RGB images(all P<0.05).Moreover,receiver operator characteristic(ROC)analysis revealed that a greater area under the curve(AUC),higher sensitivity and specificity in evaluation of HER2 gene were achieved by MS images(AUC = 0.91,89.1%,83.2%)than RGB images(AUC = 0.87,84.5%,81.8%).There was no significant difference between quantitative results of RGB images and clinico-pathological characteristics(P>0.05).However,by the analysis of MS images,the total signal OD values of HER2 positive expression were correlated with lymph node status(P = 0.02)and histological grades(P = 0.04).Additionally,the consistency test results indicated the inter-observer agreement was more robust in MS images for HER2(interclass correlation coefficient(ICC)= 0.95,rs=0.94),CK5/6(ICC = 0.90,rs = 0.88)and ER(ICC =0.94,rs =0.94)(all P<0.001)than that in RGB images for HER2(ICC = 0.91,rs = 0.89),CK5/6(ICC = 0.85,rs = 0.84)and ER(ICC = 0.90,rs = 0.89)(all P<0.001).Conclusion This study suggested that the application of MS images in quantitative IHC analysis of invasive BC could obtain higher accuracy,reliability,and more information of protein expression in relation to clinico-pathological characteristics than that of conventional RGB images.It may become an optimal IHC digital imaging system used in quantitative pathology.Part 2.A comparative performance analysis of multispectral and RGB imaging on HER2 status evaluation for the prediction of breast cancer prognosisBackground Despite the extensive application of MSI in biomedical multidisciplinary researches,there is a paucity of data available regarding the implication of MSI in tumor prognosis prediction.We compared the behaviors of MS and conventional RGB images on assessment of HER2 immunohistochemistry to explore their impact on outcome in patients with invasive BC.Methods Tissue microarrays(TMAs)containing 240 BC patients were introduced to assess HER2 status by IHC staining and to capture the corresponding RGB and MS images acquired by traditional RGB and MSI systems.The total and average signal OD values of the positive regions of HER2 were quantitatively analyzed using CRi Nuance image analysis software.The prognostic values of the quantitative results of MSI and RGB imaging methods on 5-year disease-free survival(5-DFS)were compared.Results HER2 bright-field MS images presented better image qualities compared with RGB images,making it easy for positive signal separation and quantification.Both the total and average signal OD values of HER2 MS and RGB images were calculated.According to the best P value principle,all patients were divided into two groups based on the different 5-DFS by X-tile software.The quantification of HER2 MS images was negatively correlated with 5-DFS in lymph node-negative and-positive patients(P<0.05),but RGB images has no statistical significance in lymph node-negative patients(P = 0.101).Multivariate analysis indicated that MS images could screen out more independent prognostic factors compared with RGB images,and the risk ratio was higher than that of RGB images.Additionally,area under curve(AUC)by ROC analysis for HER2 MS(AUC = 0.649;95%CI,0.577-0.722)was greater than that for HER2 RGB(AUC = 0.596;95%CI,0.522-0.670)and clinical stage(AUC-0.642;95%CI,0.571-0.713),but lower than N stage(AUC = 0.696;95%CI,0.628-0.763)and histological grade(AUC = 0.748,95%CI,0.686-0.811)in predicting the risk for recurrence.More importantly,the quantification of HER2 MS images has higher prediction accuracy than that of HER2 RGB images(69.6%vs.65.0%)on 5-DFS.Conclusions This study indicated that better information on BC prognosis could be obtained from the quantification of HER2 MS images and MS images might perform better in predicting invasive BC prognosis than conventional RGB images.Part 3.Colorectal cancer histopathology image analysis:A comparative study of prognostic values of automatically extracted morphometric nuclear features in multispectral and RGB imageryBackground MSI has emerged as a powerful tool for cancer detection and classification.The lack of clear consensus over the utility of MS bright-field imaging prompted our team to explore the benefit of using MSI on the prediction of CRC prognosis.We compared the behaviors of MS and standard RGB images on quantitative assessment of hematoxylin-eosin(HE)-stained histopathological images using computer-aided image(CAI)analysis to investigate their impact on outcome in patients with CRC.Methods We used a pixel-wise support vector machine(SVM)classifier for gland-stroma segmentation of HE-stained MS and RGB images of CRC paraffin blocks,and a marker-controlled watershed algorithm for nuclei segmentation of different images.The relationship between all extracted morphologic parameters and patients’ 5-DFS was studied and compared.Results Forty-seven morphological nuclear features were extracted after segmentation,and 8 parameters derived from MS images identified by Kaplan-Meier analysis were significantly associated with 5-DFS(P<0.05),but only 7 parameters in RGB images(P<0.05).The higher accuracy,precision,and Dice index in pixel-based nuclei segmentation for MS images were achieved than that in RGB images(0.879 vs.0.861;0.686 vs.0.667;0.634 vs.0.610).Additionally,four nagetive morphological parameters(number,circularity,eccentricity and major axis length)with area under curve(AUC)>0.5 and 4 positive mathematical parameters(average area,average perimeter,total area/total perimeter),average(area/perimeter))with AUC<0.5 in ROC analysis were combined into integrated parameter 1 and integrated parameter 2,respectively.The same integrated method was applied to the analysis results of RGB images.Multivariate Cox proportional hazards model indicated that integrated parameter 1(P<0.001)and 2(P<0.001)in MS images as an independent prognostic factor for 5-DFS,but only integrated parameter 1(P<0.001)in RGB images.More importantly,the quantification of HE MS images has higher prediction accuracy than that of HE RGB images(73.1%vs.70.1%)on 5-DFS of CRC patients.Conclusions This study demonstrated that quantitative evaluation of HE-stained MS image could yield more information and better predictive performance on CRC prognosis versus conventional RGB images by CAI analysis,thereby contributing to the development of precision oncology.