A framework for distinguishing benign from malignant breast histopathological images using deep residual networks

Author(s):  
Ziba Gandomkar ◽  
Patrick C. Brennan ◽  
Claudia Mello-Thoms
2018 ◽  
Vol 11 (3) ◽  
pp. 1335-1343
Author(s):  
P.M. Shivamurthy ◽  
T.N. Nagabhushan ◽  
Vijaya Basavaraj

Nuclear pleomorphism is considered to be one of the most significant shape based feature adapted in grading the cancer through the pathological studies of the H&E stained tissue slides. Microscopic study of manually extracting the feature is highly laborious and misleads the pathologists during grading. Digitization of the slides has given rise to various segmentation approaches to extract the nuclei shape to assess the degree of pleomorphism. Here, a novel approach of initializing and evolving the distance regularized level sets (DRLS) for the detection and segmentation of the nuclei has been presented. In this work, two major objectives have been achieved. First, a novel geometric approach has been devised for the detection of centroids of each nuclei in the occluded region and second, a shape prior model has been presented for the extraction of gradient information through morphological operations. The multiple level set implementation of the DRLS contours are initialized using the centroids detected and driven through the gradient computed. The proposed method has been experimented over the images of benign and malignant breast cancer tissue obtained from BeakHis dataset. A quantitative analysis of the results have shown that a 97% of object detection accuracy and 78% of overlap resolution has been achieved through the proposed model. A comparative study with that of geodesic active contours have indicated an improvement in the segmentation accuracy measure of 9-10 pixel difference.


1997 ◽  
Vol 36 (08) ◽  
pp. 282-288 ◽  
Author(s):  
T. Atasever ◽  
A. Özdemir ◽  
I. Öznur ◽  
N. I. Karabacak ◽  
N. Gökçora ◽  
...  

Summary Aim: Our goal was to determine the clinical usefulness of TI-201 to identify breast cancer in patients with suspicious breast lesions on clinical examination, and/or abnormal radiologic (mammography and/or ultrasonography) findings. Methods: TI-201 scintigraphy were performed in sixty-eight patients with 70 breast abnormalities (51 palpable, 19 nonpalpable) and compared with mammography and ultrasonography (US). Early (15 min) and late (3 h) images of the breasts were obtained following the injection of 111 MBq (3 mCi) of TI-201. Visual and semiquantitative interpretation was performed. Results: Final diagnosis confirmed 52 malignant breast lesions and 18 benign conditions. TI-201 visualized 47 of 52 (90%) overall malignant lesions. Thirty-eight of 40 (95%) palpable and 9 of 12 (75%) nonpalpable breast cancers were detected by TI-201 scintigraphy. The smallest mass lesion detected by TI-201 measured 1.5x1.0 cm. Eleven breast lesions were interpreted as indeterminate by mammography and/or sonography. TI-201 scintigraphy excluded malignancy in 7 of 8 (88%) patients with benign breast lesions interpreted as indeterminate. Five of the 18 (28%) benign breast lesions showed TI-201 uptake. None of the fibroadenoma and fibrocystic changes accumulated TI-201. TI-201 scintigraphy, mammography and ultrasonography showed 90%, 92%, 85% overall sensitivity and 72%, 56%, 61% overall specificity respectively. Twenty-one of the 28 (75%) axillary nodal metastatic sites were also detected by TI-201. In malignant and benign lesions, early and late lesion/contralateral normal side (L/N) ratios were 1.58 ± 0.38 (mean ± SD) and 1.48 ± 0.32 (p >0.05), 1.87 ± 0.65 and 1.34 ± 0.20 (p<0.05) respectively. The mean early and late L/N ratios of malignant and benign groups did not show statistical difference (p>0.05). Conclusion: Overall, TI-201 scintigraphy was the most specific of the three methods and yielded favourable results in palpable breast cancers, while it showed lower sensitivity in nonpalpable cancers and axillary metastases. Combined use of TI-201 scintigraphy with mammography and US seems to be useful in difficult cases, such as dense breasts and indeterminate breast lesions.


1994 ◽  
Vol 71 (05) ◽  
pp. 684-691 ◽  
Author(s):  
László Damjanovich ◽  
Csaba Turzó ◽  
Róza Ádány

SummaryThe plasminogen activation system is a delicately balanced assembly of enzymes which seems to have primary influence on tumour progression. The conversion of plasminogen into serine protease plasmin with fibrinolytic activity depends on the actual balance between plasminogen activators (urokinase type; u-PA and tissue type; t-PA) and their inhibitors (type 1 and 2 plasminogen activator inhibitors; PAI-1 and PAI-2). The purpose of this study was to determine the exact histological localization of all the major factors involved in plasminogen activation, and activation inhibition (plasmin system) in benign and malignant breast tumour samples. Our results show that factors of the plasmin system are present both in benign and malignant tumours. Cancer cells strongly labelled for both u-PA and t-PA, but epithelial cells of fibroadenoma samples were also stained for plasminogen activators at least as intensively as tumour cells in cancerous tissues. In fibroadenomas, all the epithelial cells were labelled for PAM. Staining became sporadic in malignant tumours, cells located at the periphery of tumour cell clusters regularly did not show reaction for PAI-1. In the benign tumour samples the perialveolar connective tissue stroma contained a lot of PAI-1 positive cells, showing characteristics of fibroblasts; but their number was strongly decreased in the stroma of malignant tumours. These findings indicate that the higher level of u-PA antigen, detected in malignant breast tumour samples by biochemical techniques, does not necessarily indicate increased u-PA production by tumour cells but it might be owing to the increased number of cells producing u-PA as well. In malignant tumours PAI-1 seems to be decreased in the frontage of malignant cell invasion; i.e. malignant cells at the host/tumour interface do not express PAI-1 in morphologically detectable quantity and in the peritumoural connective tissue the number of fibroblasts containing PAI-1 is also decreased.


2020 ◽  
Vol 23 (1) ◽  
pp. 10-12
Author(s):  
Nelema Jahan ◽  
Md. Mamunur Rahman ◽  
Mohammad Shahidul Alam ◽  
Md Saiful Islam

Background: A breast lump is the most common symptom associated with both benign and malignant breast diseases. Therefore, a distinction of benign from malignant lump is of importance for proper management. Though a definitive diagnosis is possible with imaging for all the lesions, histopathological study is proven essential for confirming the diagnosis. Objective: The objective of this study was to evaluate the role of USG and histopathological findings of different breast lump in diagnosis and their comparison. Methods: A prospective study was conducted over a period of one year from January 2017 to December 2017. A total of 116 patients were included in this study. All breast lumps underwent surgery and the ultrasound findings of these lumps were compared with the histopathological findings. Data were collected from these patients by a preformed questionnaire and finally the data were analyzed. Results: Out of 116 patients only 21 cases were reported as malignant in ultrasound report but histopathology revealed 31 malignant patients. On histopathological examination 10 benign cases turned out to be malignant. Conclusion: The present study was undertaken to evaluate in diagnosing breast mass lesions individually by ultrasound and compared with histopathology for definitive management of a patient. Journal of Surgical Sciences (2019) Vol. 23 (1) : 10-12


Sign in / Sign up

Export Citation Format

Share Document