[Show abstract][Hide abstract]ABSTRACT: Variations have been shown to exist in the diagnosis of cancers by different pathologists. We report on an experiment in observation of cancer features. The accuracy of the participants diagnoses showed an expected improvement when plotted against experience. When the number of features reported was plotted against experience the result showed a variation by specialists and those with less experience. The results have provided a platform for the lexicon required in the development of an Image Description Language (IDL) for pathologists to aid in reducing the variations in reporting.
No preview · Article · Jan 2010 · International Journal of Biomedical Engineering and Technology
[Show abstract][Hide abstract]ABSTRACT: Analysis of tissue is essential in dealing with a number of problems in cancer research. The identification of normal, dysplastic and cancerous colonic mucosa is an example of such a problem. In this paper, texture analysis techniques have been employed with the purpose of measuring characteristics of the tissue images. Those include histogram, grey-level difference statistics and co-occurrence matrix feature extraction algorithms. These characteristics are used as inputs for two different artificial intelligence approaches to address the image classification problem; a genetic algorithm and an artificial neural network. No significant differences have been found in the classifications obtained by both methodologies.
[Show abstract][Hide abstract]ABSTRACT: Patients with an intrathoracic oesophagogastrostomy after subtotal oesophagectomy experience profound duodenogastro-oesophageal reflux (DGOR). This study investigated the degree of mucosal injury and histopathological changes in oesophageal squamous epithelium after subtotal oesophagectomy with gastric interposition in relation to the extent of postoperative DGOR.
Serial endoscopic assessment and systematic biopsy at the oesophagogastric anastomosis was undertaken in 40 patients following curative radical subtotal oesophagectomy and reconstruction with a gastric conduit subjected to a pyloroplasty. Thirty patients subsequently underwent combined 24-h ambulatory pH and bilirubin monitoring.
Grade I-III oesophagitis was identified in 14 patients and oesophageal columnar epithelium in 19 patients. Biopsies from columnar regeneration revealed cardiac-type epithelium in ten patients and intestinal metaplasia in nine. Seven patients followed serially showed progression from cardiac-type epithelium to intestinal metaplasia. The incidence of Barrett's metaplasia was similar irrespective of the histological subtype of the resected tumour. Patients with oesophageal columnar epithelium had significantly higher acid (P = 0.015) and bilirubin (P = 0.011) reflux.
Severe DGOR occurs following subtotal oesophagectomy and provides an environment for the acquisition of Barrett's metaplasia via a sequence of cardiac epithelium and eventual intestinal metaplasia.
No preview · Article · Sep 2003 · British Journal of Surgery
[Show abstract][Hide abstract]ABSTRACT: Pancreatic cancer is frequently associated with intense growth of fibrous tissue at the periphery of tumours, but the histopathological quantification of this stromal reaction has not yet been used as a prognostic factor because of the difficulty of obtaining quantitative measures using manual methods. Manual histological grading is a poor indicator of outcome in this type of cancer and there is a clinical need to establish a more sensitive indicator. Recent pancreatic tumour biology research has focused upon the stromal reaction and there is an indication that its histopathological quantification may lead to a new prognostic indicator.
Histological samples from 21 cases of pancreatic carcinoma were stained using the sirius red, light-green method. Multiple images from the centre and periphery of each tumour were automatically segmented using colour cluster analysis to subdivide each image into representative colours. These were classified manually as stroma, cell cytoplasm or lumen in order to measure the area of each component in each image. Measured areas were analysed to determine whether the technique could detect spatial differences in the area of each tissue component over all samples, and within individual samples.
Over all 21 cases, the area of stromal tissue at the periphery of the tumours exceeded that at the centre by an average of 10.0 percentage points (P < 0.001). Within individual tumours, the algorithm was able to detect significantly more stroma (P < 0.05) at the periphery than the centre in 11 cases, whilst none of the remaining cases had significantly more stromal tissue at the centre than the periphery.
The results demonstrate that semi-automated analysis can be used to detect spatial differences in the area of fibrous tissue in routinely stained sections of pancreatic cancer.
No preview · Article · Jul 2003 · Physics in Medicine and Biology
[Show abstract][Hide abstract]ABSTRACT: Analysis of tissue using image processing techniques is essential for dealing with a number of problems in cancer research. The identification of normal and cancerous colonic mucosa is such a problem. In this paper texture analysis techniques are used to measure certain characteristics of normal and cancerous tissue images. A genetic algorithm undertakes the analysis of those results in order to determine the operations useful for the given problem and in the most appropriate operation combination for the purpose of maximising the classification accuracy. The system developed for undertaking those tasks has been implemented on a cluster of Linux workstations using distributed computing techniques. A distributed programming message-passing library, PVM (Parallel Virtual Machine), provides the basis for building this system.
[Show abstract][Hide abstract]ABSTRACT: Objective measurements of tissue area during histological examination of carcinoma can yield valuable prognostic information. However, such measurements are not made routinely because the current manual approach is time consuming and subject to large statistical sampling error. In this paper, a semi-automated image analysis method for measuring tissue area in histological samples is applied to the measurement of stromal tissue, cell cytoplasm and lumen in samples of pancreatic carcinoma and compared with the standard manual point counting method.
Histological samples from 26 cases of pancreatic carcinoma were stained using the sirius red, light-green method. Images from each sample were captured using two magnifications. Image segmentation based on colour cluster analysis was used to subdivide each image into representative colours which were classified manually into one of three tissue components. Area measurements made using this technique were compared to corresponding manual measurements and used to establish the comparative accuracy of the semi-automated image analysis technique, with a quality assurance study to measure the repeatability of the new technique.
For both magnifications and for each tissue component, the quality assurance study showed that the semi-automated image analysis algorithm had better repeatability than its manual equivalent. No significant bias was detected between the measurement techniques for any of the comparisons made using the 26 cases of pancreatic carcinoma. The ratio of manual to semi-automatic repeatability errors varied from 2.0 to 3.6. Point counting would need to be increased to be between 400 and 1400 points to achieve the same repeatability as for the semi-automated technique.
The results demonstrate that semi-automated image analysis is suitable for measuring tissue fractions in histological samples prepared with coloured stains and is a practical alternative to manual point counting.
Full-text · Article · May 2002 · Physics in Medicine and Biology
[Show abstract][Hide abstract]ABSTRACT: Analysis of tissue using image processing techniques is useful for dealing with a number of problems in cancer research. One such problem is the identification of normal, dysplastic and cancerous colonic mucosa. This research aims to identify the main image processing techniques necessary for this particular task, and to develop an automatic image classification system. Several texture analysis image-processing algorithms have been employed in this paper. Those can be separated into three categories, histogram features, grey-level difference statistics and co-occurrence matrix feature extraction algorithms. The classification system is implemented on a distributed system using PVM (Parallel Virtual Machine). PVM is a library for distributed application programming.
[Show abstract][Hide abstract]ABSTRACT: Co-occurrence matrices are commonly used to extract fine texture information from images, and have been found to be a useful tool for measuring dysplasia in histological images of the colon. Pathologists, however, measure dysplasia in tissue samples at structural as well as cytological levels. We present our findings after investigating modifications to the cooccurrence matrix technique to measure this low frequency colour texture information for the classification of colon cancer images.
[Show abstract][Hide abstract]ABSTRACT: Worldwide, colorectal cancer is the third most common malignant neoplasm. Automated classification of cytological images of colon tissue samples has been investigated, but diagnosis in all cases still requires human judgement. With the large numbers of cases of colon cancer each year, the workload placed on pathologists is immense. Texture is a powerful discriminating metric and the use of grey-level texture for classification of colon images has been extensively researched. One common technique is the extraction of texture metrics from grey-level co-occurrence matrices. However, using grey-scale images discards information contained in the differences of hue and saturation that may provide further classification information. We present the findings of an investigation of the discriminating ability of colour texture using co-occurrence matrices. Comparisons are made between grey-scale and colour texture analysis. Using statistical analysis, we show that classification using colour texture offers an improvement over classification based solely on grey-level texture.