Marcin Maciejewski

Level-set image processing methods in medical image segmentation

Abstract In this paper, two image processing methods for use in medical image processing based on the level set method are described. The theoretical bases are described and the methods are applied to a set of sample computed tomography images. The results are then compared. The results indicate that the Chan-Vese method is more useful for image segmentation in medicine than the distance-regulated method owing to both the significant differences in calculation time and the quality of results obtained for noisy images.

ECG parameter extraction and classification in noisy signals

The ECG acquisition procedure is one of the mostly used elements during initial patient examination upon hospital admission. It provides significant information about the circulatory system, electrolytic balance and even substance abuse. The test is quick, cheap, and safe for the patient due to the noninvasive nature. Nevertheless, the signal can vary significantly between individual people due to multiple factors, including differences in anatomical build of patients. Also, the ECG signal can include noise from multiple sources, especially when sampled using a mobile device. It is important for the classification algorithm to be robust enough to work in noisy conditions for as many cases as possible. The classification method described in this paper proceeds in several distinctive steps. The first operation is data preparation and wavelet filtering. Afterwards the QRS complexes are detected using the Pan-Tompkins method. The following steps include peak detection and polynomial approximations to calculate the positions and length of both P and T waves. The diagnostically relevant parameters are later used for classification using Naive Bayes and Support Vector Machine classifiers. The results obtained from the classification are presented for a group of over 50 patients both before and after normalized physical exercise.

Device to determine the level of peripheral blood circulation and saturation

The paper evaluated the diagnostic value of laser photoplethysmography when examining patients with chronic lower limb ischemia. A statistical analysis of the research results was made, and diagrams of relationship between the degrees of ischemia and blood flow are presented. Development of the device to determine the level of peripheral blood circulation and saturation was presented. Also additional accessories in the form of optical fibers for different applications were suggested.

Improving the quality of the ECG signal by filtering in wavelet transform domain

The article concerns the research methods of noise reduction occurring in the ECG signals. The method is based on the use of filtration in wavelet transform domain. The study was conducted on two types of signal - received during the rest of the patient and obtained during physical activity. For each of the signals 3 types of filtration were used. The study was designed to determine the effectiveness of various wavelets for de-noising signals obtained in both cases. The results confirm the suitability of the method for improving the quality of the electrocardiogram in case of both types of signals.

Prediction of acute pancreatitis in the earliest stages - role of biochemical parameters and histopathological changes.

For many years, there has been a search for a set of biochemical parameters that could facilitate the assessment of severity, prognosis, and administration of early and appropriate treatment in acute pancreatitis. Administration of treatment within the first 48 hours since admission is associated with many problems of distinguishing patients with a mild form of acute pancreatitis (AP) from those with a severe form of acute pancreatitis. STUDY AIM To assess the relationship between the extent of change in the concentration of 10 selected biochemical indicators: amylase, lipase, total bilirubin, creatinine, uric acid, aspartate transaminase, alanine transaminase, glucose, magnesium, and iron and histopathological lesions in the pancreas within 2 and 6 hours since induction of AP. The selected time periods correspond to the first and the second day of the disease in people, respectively. MATERIAL AND METHODS The experiments were conducted in 110 male Wistar rats weighing from 250 to 300 g. Experimental animals were divided into three groups: Z - a group in which the ranges of the studied factors and histological structure were established; K - a group of animals operated on which were injected with 0.9% NaCl into the biliary-pancreatic duct; E - a group of animals operated on in which acute pancreatitis was induced by an injection of 5% sodium taurocholate into the biliary-pancreatic duct. Animals from the K and E groups were randomly assigned to one of five subgroups from which the material for biochemical and histological examinations was collected at 2 h and 6 h since the induction of AP. Whole pancreases were dissected for histological examinations, and the samples were dyed with hematoxylin and saturated alcoholic eosin solution. The degree of pancreatic lesions was assessed according to the Spormann score. Quantitative variables were characterized by arithmetic means, standard deviations, medians, minimum and maximum values, and 95% CIs. RESULTS In histological preparations from rats from the E group, after 2 hours, edematous lesions, neutrophilic infiltrations in the pancreatic parenchyma, together with single petechiae started to appear and were observed. After 6 hours, the lesions became more intense, and minor foci of coagulation necrosis and minor foci of purulent inflammation in the fatty tissue appeared. Within 2 hours, statistically significant differences in the amount of four markers: creatinine, ALT, amylase, and magnesium were observed. After six hours, statistically significant differences in the amount of two markers: AST and glucose were seen. The correlations between histological assessments according to the Spormann scale and biochemical indicators were investigated, and it was observed that within 2 hours the intensity of pancreatitis increased together with an increase in AST. In group K, within 6 hours, the intensity of inflammatory infiltration increased together with an increase in creatinine concentration (correlation coefficient 0.95; p=0.0138). In group E, in the period of 2 hours, lesion intensity in the form of inflammatory infiltration increased together with an increase in the AST level (correlation coefficient 0.90; p=0.0063) and an increase in the iron level (correlation coefficient 0.78; p=0.0399). In the same group and in the same period, an increase in the AST level (correlation coefficient 0.79; p=0.0343) was associated with an increase in lesion intensity in the form of ecchymoses. Inflammatory infiltration increased (correlation coefficient -0.87; p=0.0117) within 6 hours, whereas the creatinine level decreased. Interesting results were obtained with the use of regression analysis - forward stepwise regression. In the period of 2 hours, if the creatinine level increased by 1, the intensity of lesions in acute pancreatitis decreased by 9.02, according to the Spormann score, while the other variables remained at a stable level. However, if ALT level increased by 1, the intensity of lesions in acute pancreatitis increased by 0.02, according to the Spormann score; and if the amylase level increased by 1, the intensity of lesions in acute pancreatitis increased by 0.01, according to the Spormann score, while the other variables remained at a stable level. CONCLUSIONS Histopathological lesions occurred prior to changes in laboratory test results, whereas significant correlations with Spormann scores were seen in the case of changes in AST and creatinine levels. The study results confirm the fact that diagnostics in acute pancreatitis is very difficult and requires monitoring of many laboratory parameters.

A device for conducting a dynamic modes of UIAB therapy with automatic process testing

The structure and circuit of the implemented device is using an environmentally friendly radiation sources and pulse photo stimulus modes with frequencies that correspond frequencies processes in BL to create bio resonance effects to accelerate the procedures. Proposed one of the possible hardware solutions are based on usage of dynamic irradiation modes and automatic continuous optical testing procedures UIAB and PP. The changes of the optical characteristics of BL provides the doctor continuous information on the effectiveness of the procedure on the patients condition.

Human ECG indicators for fast screening and evaluation

Telemedical system design and implementation requires numerous steps. It is necessary to evaluate the operation of algorithms responsible for analysis and detection of life-threatening situations. By performing ECG analysis it is possible to obtain information about the overall patient health condition as well as detailed information about the circulatory system condition. To achieve that goal one must gather, filter and process data. Data was gathered using a purposely built device from a group of four volunteers. Available data set was processed to obtain information about the patients condition. Pan-Tompkins algorithm was used to detect R peaks and calculate heart rhythm. Afterward the rest of parameters were extracted in time domain using windowed peak detection and polynomial estimation. The parameters were calculated as delays between appropriate points in the signal. The method proved to be able to extract parameters in some of the cases, and proved limited effectiveness in situations where physical activity was significant. It was nevertheless possible to eliminate noise from the mains, the trend and higher frequency noise Further improvements need to be introduced to increase the method’s robustness in the presence of significant muscle noise.

Human ECG signal parameters estimation during controlled physical activity

ECG signal parameters are commonly used indicators of human health condition. In most cases the patient should remain stationary during the examination to decrease the influence of muscle artifacts. During physical activity, the noise level increases significantly. The ECG signals were acquired during controlled physical activity on a stationary bicycle and during rest. Afterwards, the signals were processed using a method based on Pan-Tompkins algorithms to estimate their parameters and to test the method.