Cytopathological image analysis using deep-learning networks in microfluidic microscopy

Cytopathologic testing is one of the most critical steps in the diagnosis of diseases, including cancer. However, the task is laborious and demands skill. Associated high cost and low throughput drew considerable interest in automating the testing process. Several neural network architectures were designed to provide human expertise to machines. In this paper, we explore and propose the feasibility of using deep-learning networks for cytopathologic analysis by performing the classification of three important unlabeled, unstained leukemia cell lines (K562, MOLT, and HL60). The cell images used in the classification are captured using a low-cost, high-throughput cell imaging technique: microfluidics-based imaging flow cytometry. We demonstrate that without any conventional fine segmentation followed by explicit feature extraction, the proposed deep-learning algorithms effectively classify the coarsely localized cell lines. We show that the designed deep belief network as well as the deeply pretrained convolutional neural network outperform the conventionally used decision systems and are important in the medical domain, where the availability of labeled data is limited for training. We hope that our work enables the development of a clinically significant high-throughput microfluidic microscopy-based tool for disease screening/triaging, especially in resource-limited settings.

On the influence of spread constant in radial basis networks for electrical impedance tomography

Electrical impedance tomography (EIT) is a non-invasive imaging technique. The main task of this work is to solve a non-linear inverse problem, for which several techniques have been suggested, but none of which gives a very high degree of accuracy. This paper introduces a novel approach, based on radial basis function (RBF) artificial neural networks (ANNs), to solve this problem, and uses several ANNs to obtain the best solution to the EIT inverse problem. ANNs have the potential to directly estimate the solution of the inverse problem with a high degree of accuracy. While different radial basis neural networks do not always perform well on different problems, they usually give good results on some specific problems. This paper evidences a strong correlation between the area of the target and the spread constant of the RBF network that gives the best reconstruction. A solution to automatically estimate the size of the target and pick the best neural network directly from voltage measurements is presented, making the reconstruction process automatic. By automatically selecting the best ANN for each specific set of voltage measurements, the proposed solution gives a more accurate reconstruction of both small and large targets.