Improved prediction of therapeutic absorbed doses of radioiodine in the treatment of thyroid carcinoma

We proposed an alternative to a monoexponential model of radioiodine kinetics to obtain a more accurate estimate of absorbed doses to postsurgical thyroid remnants. We suggested that part of the difference between the predicted and the actually absorbed therapeutic doses of (131)I, usually explained by radiation damage of thyroid cells, can be attributed to errors resulting from inadequate sampling of data and oversimplified modeling.

METHODS

A standard monoexponential model and alternative biphasic model (incorporating both radioiodine uptake and clearance) were used on 2 sets of patient data to fit time-activity measurements after administration of diagnostic and therapeutic activities of radioiodine. One set of data consisted of 633 records of routine measurements, and the second set consisted of 71 prospectively collected records with measurements performed more frequently and for a longer time. The time-activity curves derived from the 2 models were used to calculate residence times for diagnostic and therapeutic activities of (131)I, and the respective residence times were compared using the paired t test. Errors of fitting and prediction of therapeutic time-activity data were also calculated.

RESULTS

With both models, a statistically significant difference (P < 0.01) was found between residence times after diagnostic administration of (131)I and residence times after therapeutic administration of (131)I. However, the effects of biphasic modeling and of improved sampling substantially reduced the difference (P < 0.01). Errors of fitting and prediction were smaller with the biphasic model than with the monoexponential model (P < 0.01).

CONCLUSION

The biphasic model more accurately predicts (131)I kinetics when applied to measurements in the short interval after diagnostic administration of radioiodine. The minimum requirement for the biphasic model is measurement twice a day at intervals > 6 h for at least 3 d after administration.

Biophysical inputs into the software "MIRDose"

Administered amount of activity decides on absorbed dose in thyroid gland during therapy of thyroid cancer tumors by 131I. Medical Internal Radiation Dose (MIRD) methodology estimates this dose as well as influence on other organs. MIRDose--the software implementation of MIRD--is permanently improving and has reached a substantial degree of maturity. Thus the reliability of the results depends predominantly on quality of the input data. The residence time and functional volume of the thyroid gland of a particular patient are the key inputs. Here we concentrate on the former one. We found that the traditionally used mono-exponential model, characterized by the effective half-life, introduces non-negligible modelling error. It cannot be improved by any data processing. For this reason, we proposed a novel accumulation model. Now we inspect influences of differences in the guessed residence time on the outputs of MIRDose. We briefly characterize MIRDose software, recall the improved model and present illustrative results of evaluations.

Experimental comparison of data transformation procedures for analysis of principal components

Results of principal component analysis depend on data scaling. Recently, based on theoretical considerations, several data transformation procedures have been suggested in order to improve the performance of principal component analysis of image data with respect to the optimum separation of signal and noise. The aim of this study was to test some of those suggestions, and to compare several procedures for data transformation in analysis of principal components experimentally. The experiment was performed with simulated data and the performance of individual procedures was compared using the non-parametric Friedman's test. The optimum scaling found was that which unifies the variance of noise in the observed images. In data with a Poisson distribution, the optimum scaling was the norm used in correspondence analysis. Scaling mainly affected the definition of the signal space. Once the dimension of the signal space was known, the differences in error of data and signal reproduction were small. The choice of data transformation depends on the amount of available prior knowledge (level of noise in individual images, number of components, etc), on the type of noise distribution (Gaussian, uniform, Poisson, other), and on the purpose of analysis (data compression, filtration, feature extraction).

Improved automatic separation of renal parenchyma and pelvis in dynamic renal scintigraphy using fuzzy regions of interest

The aim of the study was to examine the physiological relevance of factors produced by a modified procedure for factor analysis of dynamic renal studies. Factor analysis has been applied locally to subsets of dynamic renal data which were well defined in both space and time domains. Optimised factor images resulting from different subsets were used as fuzzy regions of interest (ROIs) for the extraction of time-activity curves corresponding to renal parenchyma, renal pelvis, vascular and spatially homogeneous background. The original procedure employed the factor images of renal parenchyma and pelvis resulting from an analysis of the interval between the peaks of parenchymal and pelvic curves. In an attempt to improve the separation of renal parenchyma and pelvis, new fuzzy ROIs were used. They correspond to the factor image of renal uptake obtained from the analysis of the early phase of the study, and to the factor image of the renal pelvis obtained from the outflow phase. The curves generated with the new fuzzy ROIs were compared with those of the original procedure and tested for the presence of known artefacts inconsistent with the expected physiological behaviour. Unlike with the original procedure, no such artefacts were found. The most striking difference was that the pelvic factor curves did not start from zero time of the study but exhibited a physiologically correct initial horizontal zero segment the length of which correlated closely with the minimum parenchymal transit time (r=0.79, n=46, P<0.001). The new method permits easy and reliable application of factor analysis to dynamic renal studies. Problems which remain to be solved are user-independent identification of the optimum factors and suboptimal performance of the method under extreme conditions. Our results provide additional evidence that factor analysis can extract physiologically relevant information quantitatively from dynamic scintigraphic data.

Hybrid phantoms for testing the measurement of regional dynamics in dynamic renal scintigraphy

A hybrid phantom is a synthetic dynamic scintigraphic study reconstructed from dynamic components of real clinical data. The aim of a hybrid simulation is to provide realistic reference studies with systematically varying quantitative diagnostic parameters for testing the quality of measurement of regional dynamics. The components of the phantom are extracted from a representative patient study using factor analysis. Prior to the construction of the phantom, selected components are modified in a known way or substituted with heterologous data. The phantom study is constituted from the altered components using a procedure inverse to factor analysis, and completed by the addition of residual noise. The hybrid phantom is a compromise between simulated and software phantoms. Unlike simulated phantoms, it preserves a significant part of data variability and the natural complex structure of scintigraphic images. Unlike software phantoms, it provides values for selected diagnostic parameters or magnitudes of their relative changes. A set of hybrid phantoms from a dynamic renal study and the results of a pilot clinical evaluation are presented.

Relative renal uptake and transit time measurements using functional factor images and fuzzy regions of interest

The aim of the study was a quantitative comparison of relative renal uptake and both the whole-kidney and the parenchymal transit time derived from factor analysis of image sequences and provided by standard clinical procedues. In order to extract the stable, well-interpretable factors, factor analysis was performed locally in the problem-specific time and spatial windows and the resulting factor images either evaluated directly as functional images or used as fuzzy regions of interest (ROIs) for the subsequent extraction of time-activity curves from the analysed data. The values of relative renal uptake of the left kidney measured in the functional factor images, which demonstrate the initial accumulation of activity in renal parenchyma (mean 51.0%), did not differ significantly from the values obtained by a standard method (mean 51.5%, r = 0.98, P<0.001). Whole-kidney transit time calculated using fuzzy ROI curves correlated well with the reference values (r = 0.84, P<0.001); however, both its mean value (336.5 s) and the standard deviation (151.5 s) were substantially greater than those of the values provided by a standard procedure (262.8+/-86.9 s). Parenchymal transit time calculated using ROI curves correlated better with the transit time through a wider corticomedullary region rather than through a narrow cortical region, which is decisive in a differential diagnosis of renal disorders. In general, values of transit times provided by factor analysis correlated well with those provided by reference methods but with a shift towards the higher numerical values. This may have been a consequence of a greater extent of the automatically extracted fuzzy ROIs, or of occasionally delayed accumulation in the upper calyces. Results of the study provide quantitative evidence that the factor analysis of dynamic data, even without the introduction of prior physiological information, may yield clinically relevant information. However, some basic requirements, such as sufficiently high sampling frequency and count rate, adaption of the method to a specific clinical task, and proper selection of time and spatial windows for locally performed analysis, have to be fullfilled if the method is to be successfully applied clinically.

Quantification of intensity variations in functional MR images using rotated principal components

In functional MRI (fMRI), the changes in cerebral haemodynamics related to stimulated neural brain activity are measured using standard clinical MR equipment. Small intensity variations in fMRI data have to be detected and distinguished from non-neural effects by careful image analysis. Based on multivariate statistics we describe an algorithm involving oblique rotation of the most significant principal components for an estimation of the temporal and spatial distribution of the stimulated neural activity over the whole image matrix. This algorithm takes advantage of strong local signal variations. A mathematical phantom was designed to generate simulated data for the evaluation of the method. In simulation experiments, the potential of the method to quantify small intensity changes, especially when processing data sets containing multiple sources of signal variations, was demonstrated. In vivo fMRI data collected in both visual and motor stimulation experiments were analysed, showing a proper location of the activated cortical regions within well known neural centres and an accurate extraction of the activation time profile. The suggested method yields accurate absolute quantification of in vivo brain activity without the need of extensive prior knowledge and user interaction.

[Clinical applications of factor analysis in nuclear medicine]

Factor analysis allows to break up dynamic radionuclide studies into constituent parts corresponding to individual compartments of the distribution space of the radiopharmaceutical in the body. Compartment structure is reflected by factor image and its dynamics is shown by factor time activity curves. The method overcomes the drawbacks of other data processing methods in dynamic scintigraphy by its objectiveness and by distinguishing the projection overlap of tissues with different dynamics. The demands on both the patient and personnel are the same as in standard procedures, yet this method extracts more information from the results of the examination. Higher diagnostic accuracy was reported especially in nuclear cardiology and nephrology. A short survey of clinical applications of factor analysis reported over the years 1982-1988 is presented.

On the existence of an unambiguous solution in factor analysis of dynamic studies

Achievement of an unambiguous solution in factor analysis of dynamic radionuclide studies depends on constraints reflecting the known properties of factors. The constraints should be tight enough to prevent ambiguity but sufficiently general in order to ensure the data-based derivation of factors. In dynamic scintigraphy, the non-negativity of factors is their essential property which is implied by the physical nature of measured quantities. Considering factors as the images of compartments in the distribution space of a radiopharmaceutical (i.e. performing the factor analysis in the spatial domain), a powerful additional constraint can be applied. This constraint is based on the presence of segments in the image matrix where the subtotal number of compartments is projected. Using this constraint, the existence of physiologically related unique solution in factor analysis can be proved providing the number of factors is chosen properly.

Rotation to simple structure in factor analysis of dynamic radionuclide studies

The physiological interpretation of factors in the factor analysis of dynamic radionuclide studies is dependent on the proper solution of the problem of factor rotation. A new solution suitable for scintigraphic data is suggested using a generalised concept of simple structure in data configurations. The method is described in detail, its efficiency is demonstrated on a phantom and its relationship to existing methods is discussed.

Enhancement of physiological factors in factor analysis of dynamic studies

Factor analysis of dynamic radionuclide studies provides their decomposition into the images and time-activity curves corresponding to the underlying dynamic structures. The method is based on the analysis of study variance and on the subsequent differential imaging of its principal components into a simplified factor space. By changing the amount and the composition of the variance processed in the analysis it is possible to enhance the factors that are important for diagnosis while the less important factors can be suppressed. In our report, a short theoretical review of the problem is given and illustrated by the analysis of dynamic cholescintigraphy. It is shown that a suitable choice of region and/or the temporal interval of interest enables the differential evaluation of such intrahepatic compartments, which could not be observed without enhancement.

Phagocytic activity of peripheral blood leukocytes in patients with Hodgkin's disease

On the basis of a simple examination of the phagocytic activity of leukocytes of the peripheral blood., with the help of the yeast cells Saccharomyces cerevisiae, we determined the functional ability of leukocytes in untreated patients with the Hodgkin's disease, in the course of the primary chemotherapy, during the complete remission and in the relapse of the disease. In all the periods we found only changes in the absolute number of phagocytic cells, corresponding to the changes in the leukocyte count (increased numbers in untreated patients and in relapse, decreased number in the course of the chemotherapy). The value of the phagocytic and activity indices in untreated patients were only moderately reduced. In the course of chemotherapy there was a more remarkable drop of the phagocytic activity only in the first half of the treatment series. In the complete remission the phagocytic activity values were normal. We did not observe significant differences between clinical stages of the Hodgkin's disease in any group. Our results correspond to the clinical course of the disease (absence of infectious complications) in the patients followed. The presented conclusions do not concern bactericidal ability of leukocytes, which is not reflected by the test method used. The test described here does not provide more data for monitoring the clinical course and chemotherapy of Hodgkin's disease than do routine examinations of the erythrocyte sedimentation and white blood count.

Individual changes of DNA catabolite excretion in the course of antitumor therapy of Hodgkin's disease

In patients with morbus Hodgkin, treated primarily by the actino- and chemotherapy, the excretion was followed of DNA catabolites (deoxycytidine, deoxyuridine, thymidine and their sum) in the course of the therapy. The dynamics was studied of changes in the time interval of interest and attention was paid to its relation to the clinical and histological type of disease and to the successful character of the therapy defined by reaching a complete remission. The group of patients as a whole was characterized by an increased excretion of catabolites in the time interval of interest. No dependence was demonstrated between the catabolite excretion and extent of the disease similarly as between the excretion and successful character of the therapy. The dynamics of the changes in the time intervals of interest was neither remarkably nor continuously increased or decreased. The test of the excretion of pyrimidine deoxyribonucleosides possesses sufficient sensitivity for demonstrating laws in relation to the therapy during group evaluation. With respect to individual variability of values of particular patients and to the absence of the relations mentioned above the test is not suitable to indicate the individual response to the anticancer therapy.