Detection of Malignancy Associated Changes in Cervical Cell Nuclei Using Feed‐Forward Neural Networks

Abstract

Normal cells in the presence of a precancerous lesion undergo subtle changes of their DNA distribution when observed by visible microscopy. These changes have been termed Malignancy Associated Changes (MACs). Using statistical models such as neural networks and discriminant functions it is possible to design classifiers that can separate these objects from truly normal cells. The correct classification rate using feed‐forward neural networks is compared to linear discriminant analysis when applied to detecting MACs. Classifiers were designed using 53 nuclear features calculated from images for each of 25,360 normal appearing cells taken from 344 slides diagnosed as normal or containing severe dysplasia. A linear discriminant function achieved a correct classification rate of 61.6% on the test data while neural networks scored as high as 72.5% on a cell‐by‐cell basis. The cell classifiers were applied to a library of 93,494 cells from 395 slides, and the results were jackknifed using a single slide feature. The discriminant function achieved a correct classification rate of 67.6% while the neural networks managed as high as 76.2%.

Advances in molecular labeling, high throughput imaging and machine intelligence portend powerful functional cellular biochemistry tools

“…High-level image interpretation has been used in a wide range of applications [Hudson and Cohen, 2000;Duda et al, 2001] including microscope images of: cervical smears [Kemp et al, 1997;Mackin et al, 1998;Van der Laak et al, 2002]; premalignant prostate, colon and esophageal tissue [Weyn et al, 2000]; and cultured cells [Boland et al, 1998;Boland and Murphy, 2001]. However, automated image interpretation using classical clinical stains (e.g., hematoxylin and eosin, and variations of the Pap stain) has only shown limited success [Bartels and Vooijs, 1999].…”

“…This suggests similarity in odd shapes occurring both in 'doublets' and in 'mitotic cells. ' The resulting performances of the automated high-level image interpreters can vary widely (anywhere between 60 and 100%) depending on how similar the object classes are, how discriminating the molecular biomarker(s) used for imaging is (are), and whether the differences in the imaged objects are simple on-off decisions, like for rare event detection of fetal nucleated red blood cells in maternal peripheral blood [Bianchi, 1999;Bajaj et al, 2000;Bianchi et al, 2002;Yamanishi et al, 2002], images of which are shown in Figure 7 (left), or fine differences, e.g., in texture of the molecular marker that can classify cells with malignancy associated changes [Kemp et al, 1997]. An example of a breast cancer specific marker, anti-cytokeratin- Fig.…”

A dynamic semisupervised feedforward neural network clustering

“…An artificial neural network (ANN) is inspired by the manner of biological nervous system, such as human brain to process the data, and is one of the numerous algorithms used in machine learning and data mining (Dasarthy, 1990; Kemp et al, 1997; Goebel & Gruenwald, 1999; Hegland, 2003; Kantardzic, 2011). In a feedforward neural network (FFNN), data processing has only one forward direction from the input layer to the output layer without any backward loop or feedback connection (Bose & Liang, 1996; McCloskey, 2000; Andonie & Kovalerchuk, 2007; Kantardzic, 2011).…”

A Pilot Study on Image Analysis Techniques for Extracting Early Uterine Cervix Cancer Cell Features

“…It takes years for an abnormal cell to grow, and we need the earliest possible sign of abnormality to let for the earliest treatment [1][2][3]. ThinPrep Pap smear screening can be useful as an early diagnostic process which can diagnose the cervical cancer [4,5]. Finding abnormal cells in ThinPrep Pap smears is an error-prone and difficult problem for pathologists.…”