Volume 28, Issue 6, June 2001
Index of content:
- PH. D. THESES ABSTRACTS
Fast algorithm for iterative image reconstruction with a small animal positron emission tomograph YAP-PET (in English)28(2001); http://dx.doi.org/10.1118/1.1376642View Description Hide Description
A fast deconvolutive expectation maximization algorithm for iterative image reconstruction was developed for use with a small animalpositron emission tomograph YAP-PET built by the Medical Physics Group at Ferrara University. It has a FOV of 4 cm diam×4 cm axially. In PET modality, the tomograph has a sensitivity of 640 cps/μCi at its center and a spatial resolution of 1.8 mm FWHM. In SPECT, the sensitivity is 2.1 cps/μCi for each detector head, with a spatial resolution of 3.5 mm FWHM. The 3D-FBP algorithm, usually used for image reconstruction, has a limited angle due to the geometry of the tomograph, which causes a serious loss in sensitivity. We developed an iterative method using the symmetries and the sparse properties of the “probability matrix,” which correlates the emission from each voxel to the detector within a coincidence tube, which we calculated before the reconstruction. We stored only a fraction of matrix elements on disk, thus avoiding on-line computation, that needs to be performed for the projection and the backprojection steps. This reduces the required storage space by a factor of and the time per iteration from hours to minutes or tens of minutes. A deconvolutive model differentiates the voxels according to their depth in a coincidence tube. This algorithm is used for both PET and SPECT data. It improves the signal-to-noise ratio and the spatial resolution. Additionally it allows quantitative data to be available on images in the SPECT mode for cardiac perfusion and uptake studies using 99m-Tc labeled radiopharmaceuticals.
Evaluation of osteoporosis using conventional radiographic methods and dual energy x-ray absorptiometry28(2001); http://dx.doi.org/10.1118/1.1376643View Description Hide Description
In India, conventional radiographs play a major part in evaluating osteoporosis since advanced bone densitometers are expensive and not widely available. Our aim in this study was to investigate the usefulness of conventional radiographic methods, viz., clavicle and metacarpal radiogrammetry, quantitative vertebral morphometry, Singh’s index, hip geometry compared to a sophisticated method, viz., DXA for studying osteoporosis in south Indian men (n=663) and women (n=741). This study provides data on BMD of the proximal femur in normal south Indian females. In women, aged above 65 years, BMD of femoral neck, trochanter, and Ward’s triangle decreased by 0.90%, 0.84%, and 1.66% per annum, respectively. In 45 pre- and post-menopausal women without fractures, total hip BMD correlated significantly ( p<0.001) with Singh’s index medial cortical thickness of the femoral shaft and neck combined cortical thickness (CCT) and %CCT of the clavicle each), and the second metacarpal and respectively). On comparing variables studied for osteoporotic women with corresponding values in the premenopausal group, the average difference in SD from the mean was greatest for clavicle radiogrammetry (−3.7 SD for %CCT of the clavicle). An empirical formula for predicting total hip BMD with good sensitivity was derived from a multiple linear regressionequation involving three independent variables, viz., CCT, %CCT of the clavicle (measured from the chest radiograph) and the patient’s age. This author’s equation, with modified weightings, has 82% sensitivity and 94% specificity and a positive and negative predictive value of 88% and 91% respectively. A chest radiograph in combination with the formula would serve as a readily available, inexpensive tool for assessing post-menopausal osteoporosis, especially in developing countries.
28(2001); http://dx.doi.org/10.1118/1.1376644View Description Hide Description
Theoretical modeling of the biological response is considered for inclusion into treatment planning algorithms. However, the results are model sensitive and therefore the choice of the model is very important. In this thesis we aimed to compare the predictions in the clinical dose range for the linear quadratic (LQ) model and the linear quadratic model with inducible repair (LQ/IR) and to determine the possible therapeutic gain of different treatment strategies, taking into consideration the microenvironmental conditions known to exist in tumors. The main focus was on the availability of oxygen and other nutrients to tumor cells, knowing that tumor vasculature is very poor compared to that in normal tissues. The brief interruption of oxygen supply to the acutely hypoxic cells determines an increased radioresistance, while the lack of oxygen and other nutrients in starved hypoxic cells results in a radiosensitization due to the reduction of cellular energy charge. It is the first time this latter aspect of tumor sensitivity was considered for a theoretical model. The modeling showed that there are some important differences in the clinical dose range between the predicted responses with the two models. Some of these differences provide an alternative explanation to the success of hyperfractionation and also explain some unusual results reported in the literature with respect to hypoxic protection. The analysis of the complex relationship between the induction of repair and the intrinsic radioresistance indicates the possible therapeutic gain that can be expected from more extreme fractionation. The existence of starved chronically hypoxic cells in tumors, with their incapacity to activate repair provides a better understanding of the tumor response to radiation treatments.
28(2001); http://dx.doi.org/10.1118/1.1376645View Description Hide Description
Automating mass chest screening for tuberculosis (TB) requires segmentation and texture analysis in chest radiographs. Several rule-based schemes, pixel classifiers, and active shape model techniques for segmenting lung fields in chest radiographs are described and compared. An improved version of the active shape model segmentation technique, originally developed by Cootes and Taylor from Manchester University, UK, is described that uses optimal local features to steer the segmentation process and outperforms the original method in segmentation tasks for several types of medical images: chest radiographs and slices from MRI brain data. In order to segment the posterior ribs in PA chest radiographs, a statistical model of the complete rib cage is constructed using principal components analysis and a method is described to fit this model to input images automatically. For texture analysis, an extension is proposed to the framework of locally orderless images, a multiscale description of local histograms recently proposed by Koenderink and Van Doorn from Utrecht University, The Netherlands. The segmentation and texture analysis techniques are combined into a single method that automatically detects textural abnormalities in chest radiographs and estimates the probability that an image contains abnormalities. The method was evaluated on two databases. On a 200 case database of clinical chest films with interstitial disease from the University of Chicago, excellent results are obtained (area under the ROC curve The results for a 600 case database from a TB screening program are encouraging (area under the ROC curve