China 
Africa 
Explore chapters and articles related to this topic
Robust Nuclei Segmentation using Statistical Level Set Method with Topology Preserving Constraint
Published in Ayman El-Baz, Jasjit S. Suri, Level Set Method in Medical Imaging Segmentation, 2019
Shaghayegh Taheri, Thomas Fevens, Tien D. Bui
Pathology is a medical specialty which concerns laboratory examination of cells and tissue samples with the purpose of diagnosis and characterization of diseases. More specifically, cytopathological and histopathological examinations of a biopsy or surgical specimen are two main branches of anatomical pathology that are commonly applied to diagnose various diseases, including cancer. Cytopathology (or cytology) refers to the microscopic investigation of samples at the cellular level and is mainly advantageous when quick preparation, staining, and interpretation procedures are needed. Despite the fact that cytopathological imagery are highly beneficial as they provide great cellular detail at low cost, cytopathological examinations alone are not sufficient for accurate diagnosis purposes. For instance, they cannot indicate whether the cancer cells are spreading into and damaging surrounding tissues. Therefore, to obtain higher diagnostic accuracy, the preliminary cytopathological tests must be confirmed by the so-called histopathological (or histological) assessments for which the overall tissue architecture is evaluated. Pathologists usually make diagnostic interferences by visual inspection of cells based on their morphological features and architecture, such as shape, position, size, number, etc. Although still being considered as the gold standard, manual examination of biological images is tedious work which requires many hours of human labor. This highlights the requirement for an automatic system that accurately measures these features in a few seconds.
Systems: packaging and applications
Published in Neil Collings, Fourier Optics in Image Processing, 2018
Cytopathology involves the visual inspection of cell tissue. It can be performed using spatial filtering in the microscope [199] or on CCD images [95] of tissue. It is an example of a repetitive task where abnormal tissue is a rare occurrence and the false negative rates are, understandably, high. Hence the interest in automating the visual inspection process. An example is cervical cancer screening, where pap-smear slides are prepared from cellular samples obtained from the cervix. Where facilities exist, such slides are sent to a cytology laboratory for staining. This is followed by high-resolution microscopy and automated feature extraction and classification of the cytoplasm and the nucleus. An optical processing approach, consisting of two stages, can be employed to isolate regions of interest (ROIs) for closer (possibly visual) inspection [180]. The first, automatic stage is the recognition of large nuclei in a hybrid correlator, using a filter which is the Fourier transform of a circular dot. Abnormal cells are identified on the basis of their enlarged nuclei. A normal cell’s nuclear diameter is between 6 and 8 μm $ \mu m $ , while an abnormal cell’s nuclear diameter ranges between 10 and 22 μm $ \mu m $ . The circular dot has a diameter which is the lower limit for the diameter of the nucleus in an abnormal cell. The correlation peak width or height can be thresholded to separate the cells with nuclear diameters less than this diameter (normal cells) from those with diameters greater than the filter diameter (abnormal cells). In information processing, simple masks, such as the FT of the circular dot, are called kernels. The convolution of an image with such a kernel is a basic technique in morphological image processing, where ‘hits’ correspond to those parts of the image that are larger than the dot. Complementary filtering can be performed on the complementary image, where ‘misses’ correspond to those parts of the complementary image that are smaller than the dot. This involves a complementary kernel to the normal image, and the combination of the two convolutions (‘hit’ and ‘miss’) is called the hit/miss transform (HMT). The HMT detects the shape/size of the cell nucleus. If the nucleus is circular with a diameter larger than 10 μm $ \mu m $ , the cell is identified as suspicious. An optoelectronic implementation of the HMT using two 256 x 256 ferroelectric-liquid-crystal SLMs detects ROIs that can be further processed by ROI classification algorithms. Where advanced facilities for cell screening (staining and high-resolution microscopy) do not exist, for example in the Third World, it has been found that the two-dimensional Fourier transform of the cell is a rich feature space which can be used to differentiate normal and abnormal cells [54].
A Systematic Review of Real-time Fine-needle Aspiration Biopsy Methods for Soft Tissues
Published in IETE Technical Review, 2022
Rahul Nadda, Ashish Kumar Sahani, Ramjee Repaka
The present study includes inspection of various patients for the presence of tumors, cancer, nodules, etc., radiographically or clinically using real-time FNA biopsy.The study considered accuracy, sensitivity, and specificity associated with real-time methods of FNA biopsy in detecting benign versus malignant tumors.Both the histopathology and cytopathology results obtained from the aspiration were reported.FNA biopsy was executed before the surgery through palpation and real-time US assistance.FNA biopsy was executed through real-time EUS assistance and CT assistance.Mean diameter of lesions and nodules.
Review of the use of telepathology for intraoperative consultation
Published in Expert Review of Medical Devices, 2018
Robin L. Dietz, Douglas J. Hartman, Liu Zheng, Clayton Wiley, Liron Pantanowitz
There are several differences between WSI and RM systems that could potentially affect the outcome of an IOC. RM allows the pathologist to focus up and down, a feature that is particularly valuable when evaluating thick or folded tissue sections, as well as cytopathology specimens or neurosurgical smears that may have 3D cell groups and/or obscuring material. Although some WSI systems have Z-stacking features to compensate for this disadvantage, the amount of time needed to scan a slide with multiple Z-stacks is impractical for an IOC. A compromise that offers the best of both worlds is a hybrid WSI system with a live mode that allows for RM use, as described above. The time required to transmit large files generated by a WSI scanner does not seem to affect the TAT versus RM systems. Evans et al. found that the average turnaround time of RM (19.98 min) was longer than that of WSI (15.68 min) [18]. This is a distinct advantage that WSI holds over RM instruments.
Segmentation of cervical nuclei using convolutional neural network for conventional cytology
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2023
Júlia Beatriz A. Teixeira, Mariana T. Rezende, Débora N. Diniz, Cláudia M. Carneiro, Eduardo J. da S. Luz, Marcone J. F. Souza, Daniela M. Ushizima, Fátima N. S. de Medeiros, Andrea G. Campos Bianchi
The limited number of cell segmentation databases interferes with the study and development of algorithms and computational tools that support medical decision-making systems in digital cytopathology since available data does not represent the routine and variability of cytology laboratories. Further data collection is required for the proper development of segmentation algorithms. Thus, this work will use and deliver a new segmentation database, CRIC Cervix-Seg, that includes nuclei and cytoplasm segmentation for real Pap smear images obtained from conventional cytology. CRIC Cervix-Seg used images from CRIC Cervix and was built using a web platform1 for the labelling process. A complete description will be available in Section 3.1.2.