Spectral and dynamic confocal fluorescence characterization of cytogenetic preparations.

Investigations were performed on fluorescent in situ hybridization (FISH) preparations to examine whether factor analysis of medical image sequences (FAMIS) can be used to isolate fluorescent probes by means of their spectral and/or extinction dynamic emission properties. FISH is used to track down chromosomes of interest in cell nuclei and mitoses. Cytogenetic techniques producing flat preparations of whole cells were assumed to preserve the probes' access to their targets. To isolate the result of hybridization in the human nuclear interphase, we used a confocal microscope. Labelling of the targets by the probes (sequences labelled by FITC and TRITC) in the nuclei stained by propidium iodide was used as a biological model. We used two methods to isolate the component parts of the model: multispectral analysis and dynamic studies. In the case of multispectral analysis, the investigation was performed on 2D and 3D sequences of 28 images obtained on a single photomultiplier (PM) detector of the confocal microscope by selection of emission through 10-nm interference filters in the range of 500-780 nm and by z-displacement in each filter setting. In the case of dynamic studies, the investigation was performed on sequences of 30-70 images obtained on the same detector by single or average integrated acquisition of 10-30 scans. Confocal scanning yields images with constant excitation time. These images were investigated by FAMIS and the results revealed that the spectra and kinetics as factors, and factor images corresponded to FITC and TRITC stained targets, as well as to propidium iodide stained interphase. In conclusion, we would verify that targets were isolated through the spectrum of the fluorescent probes and could be distinguished from the propidium iodide used to stain the nuclei. It was also possible to distinguish them from the propidium iodide by taking into account differences in photobleaching of the different fluorochromes. The study leads us to process displacements by registration methods prior to factor analysis to improve the results.

Comparison between factor analysis of dynamic structures and Fourier analysis in detection of segmental wall motion abnormalities: a clinical evaluation.

To evaluate Factor Analysis of Dynamic Structures (FADS) versus or in association with other methods, a protocol was set up including as 'gold standard' investigation the left ventricular angiography (LVA) and processing by Fourier Analysis (FA), and FADS with different variants. To refine the diagnosis of Regional Wall Motion Abnormalities (RWMA), processing was done on a sectorial basis for more accurate spatial localization and functional description. 53 patients were studied (8 normal, 45 with coronary artery disease). FADS gave better results than FA on a sectorial basis. Total agreement between FADS and LVA was obtained in 208/265 (78%), while FA was in agreement with LVA in only 167/265 segments (63%). Globally, FADS was significantly better than FA (Z-test: p < 0.05). When only the diagnosis of maximal abnormality was considered, FA and FADS are statistically equivalent. The superiority of FADS vs FA is more obvious in the diagnosis of hypokinesia. Most FA discrepancies corresponded to underestimation of WMA.

Factor analysis of medical image sequences in MR of head and neck tumors.

PURPOSE: To evaluate factor analysis of medical image sequences (FAMIS), a means whereby physiologic contrast enhancement kinetics, called factors, and their spatial distribution, termed factor images, are estimated after acquisition of dynamic MR images. The method is intended to recognize and characterize the different tissue kinetics automatically. METHODS: This method was evaluated in a series of 22 patients with head and neck tumors. Eleven patients presented with a previously untreated lesion. Six were examined for tumor recurrence, previously treated by multiple therapies. Five patients had preoperative chemotherapy and underwent MR before and after chemotherapy. In all cases, MR images were correlated with surgical and pathologic data. MR examinations were performed on a 1.5-T unit with static sequences and dynamic sequences acquired after bolus injection of gadolinium and processed by FAMIS. RESULTS: FAMIS was able to identify three factors representing contrast-enhancement kinetics and their associated factor images. The neoplastic component was associated with the earlier factor image, F1. Fibrosis and chemotherapy and/or radiation-induced changes were associated with the two later factors, F2 and F3. The limits of this method were highly vascularized tissues whose earlier factor was similar to that of neoplastic tissues (mucosae and salivary glands), patient motion, responsible for artifacts in FAMIS, and lesions of less than 5 mm. CONCLUSION: FAMIS of dynamic MR studies was useful for differentiating neoplastic tissue from tissue having undergone changes by chemotherapy and/or radiotherapy, but it did not improve the ability of MR to characterize neoplastic tissues in previously untreated patients.

Factor analysis as a means of determining response to chemotherapy in patients with osteogenic sarcoma.

The prognosis of localized osteogenic sarcoma (OS) has improved considerably since the introduction of neoadjuvant chemotherapy. However, there is a subset of patients who do not show full benefit from neoadjuvant chemotherapy because of chemoresistance. The early identification of poor responders to chemotherapy during neoadjuvant therapy remains difficult. In order to evaluate the role of bone scintigraphy we report our experience of dynamic technetium-99m hydroxymethylene diphosphonate bone scintigraphy in 19 cases of paediatric osteogenic sarcomas. Before the beginning of chemotherapy, a dynamic scan was recorded during 30 min followed by static images at 3 h. The procedure was repeated halfway through the course of chemotherapy (6th week). Histological grading of the response to chemotherapy was carried out in the 12th week, showing nine good responses and ten poor responses. Factor analysis of dynamic structures (FADS) applied to dynamic scans allowed us to identify three factors termed vascular, "soft tissue" and osseous factors. The effect of chemotherapy on each factor was evaluated. Using FADS we were able to detect all the poor histological responders with the combination of vascular and osseous factors. Six out of nine good histological responders were also classified as scintigraphic responders. FADS applied to dynamic bone scans allowed us to identify at an early stage all the poor histological responders to neoadjuvant chemotherapy. This method may have clinical relevance for the therapeutic strategy in patients with OS.

A statistical model for the determination of the optimal metric in factor analysis of medical image sequences (FAMIS).

A statistical model is added to the conventional physical model underlying factor analysis of medical image sequences (FAMIS). It allows a derivation of the optimal metric to be used for the orthogonal decomposition involved in FAMIS. The oblique analysis of FAMIS is extended to take this optimal metric into account. The case of scintigraphic image sequences is used. We derive in this case that the optimal decomposition is obtained by correspondence analysis. A scintigraphic dynamic study illustrates the practical consequences of the use of the optimal metric in FAMIS.

Extraction of functional volumes from medical dynamic volumetric data sets.

A method based on factor analysis is presented to process dynamic volumetric (t + 3D) data sets acquired for flow, excretion, or metabolic studies. It estimates a reduced number of underlying physiological kinetics and their associated spatial distributions, corresponding to functional volumes, using dedicated algorithms. The global (t + 3D) approach is shown to be superior to the conventional one, which repeats estimations on each (t + 2D) data set, obtained for each slice or projection of the volume.

Target apex-seeking in factor analysis of medical image sequences.

The aim of factor analysis of medical image sequences (FAMIS) is to estimate a limited number of physical or physiological fundamental functions. Its oblique rotation stage strongly affects the quality and the interpretation of the resulting estimates (factors and factor images). A new target apex-seeking method which integrates physical or physiological knowledge in this stage is described. This knowledge concerns some of the fundamental functions and reacts on the determination of all the factors. A simulated spectral study illustrates the method. We discuss its properties in comparison with the other approaches using a priori physical or physiological information.

FAMIS: a software package for functional feature extraction from biomedical multidimensional images.

An increased number of image sequences is acquired in all modalities of the biomedical imaging field in order to study displacement or metabolism of a tracer or a contrast agent. It requires effective processing methods to estimate the underlying physiological components. We have developed a software package based on factor analysis algorithms which can adapt to various imaging modalities and its extension to double-indexed image sequences. We describe general characteristics of the software and present the main points of the user-friendly interface. The performances of the package are discussed and the possibilities of the methodology are illustrated using an example in magnetic resonance imaging.

Diagnosis of malignancy in thyroid nodules by factor analysis of spectral and dynamic structures: a simultaneous dual-isotope dynamic study with thallium-201 and iodine-131.

The aim of this study was to identify malignant thyroid nodules using iodine-123 and thallium-201 simultaneous dynamic acquisition. The image sequences acquired were processed by factor analysis of spectral and dynamic structures (FASDS). Some 49 patients were investigated, and their diagnoses were confirmed by histological examination. Data processing enables the estimation of the spectra of the two isotopes and the evaluation of the kinetics and spatial structures related to each tracer. The superimposition of thallium and iodide sum images allowed us to delineate the nodule accurately. Two groups were defined: 21 patients who had 201Tl uptake in the nodule, and 28 who had none. In the first group, 5 nodules were carcinomas, whereas all nodules in the second group were benign. The results of the 201Tl dynamic study improved the diagnosis of carcinoma as the number of false-positive cases decreased. FASDS succeeds in extracting spectral and kinetic information, proving its usefulness in clinical diagnosis.

Processing of Xe-127 regional pulmonary ventilation by factor analysis and compartmental modeling.

A regional lung ventilation was modeled in five cases using Xe-127 during the first 3 min of wash-out. A factor analysis of the dynamic structures algorithm allowed estimation of the elementary kinetics and their respective proportions contained in time series images. Each factor was interpreted as the sampling of a compartment. It was associated to a factor image representing its spatial distribution and to a percentage of the total collected information. In the study, three factors were estimated: a fast clearance in the lower lung regions (28.9%), a slower clearance in the upper regions (33.4%), and a slow kinetics in blood and tissues containing dissolved xenon (37%). Estimates of the kinematics components obtained from a factor analysis of dynamic structures (FADSs) were used for compartmental analysis. The authors applied this method to the study of the regional ventilation distribution to establish the model and some possible variations. (FADSs) and region of interest results were used for modeling and compared.

Positive anticalcitonin immunoscintigraphy in patients with medullary thyroid carcinoma.

A cocktail of three monoclonal F(ab')2 fragments against three distinct epitopes of calcitonin or PDN 21 was labelled with either 111In or 131I. These F(ab')2 fragments, a control 125I-F(ab')2 fragment and 99mTc-pertechnetate were injected into four patients suffering from medullary thyroid carcinoma. Scintigraphy data were processed by energy factor analysis for an optimal separation of images corresponding to each isotope. The best tumor detection was obtained 1-3 days after injection of the 111In-F(ab')2 cocktail which clearly labeled the thyroid tumors in the four patients (smallest tumor detected, 0.6 cm) as well as lymph node and bone metastases. In the liver, positive detection was only successful with the 131I-labeled cocktail. These results were confirmed by counting rates of resected specimens which provided average specificity indices ranging from 3.3 to 13.1. Anticalcitonin antibodies could be particularly useful for immunoscintigraphy detection of residual or recurrent medullary thyroid carcinoma in patients with elevated calcitonin serum level.

The role of factor analysis in the evaluation of new radiopharmaceuticals.

Factor analysis of dynamic scintigraphic studies has been proposed for a variety of clinical applications. This method also called FADS (Factor Analysis of Dynamic Structures) enables spatial separation of complex images into discrete factors according to their time/activity characteristics. FADS, which does not require a priori formulation of a compartmental model of tracer kinetics, is particularly suitable for the evaluation of new radiolabeled compounds as potential radiopharmaceuticals. In animals as well as in humans it is possible to obtain information on the spatial time-distribution of tracers by analyzing computer acquired scintigraphic studies. On the basis of data obtained and analyzed with this method using [123I]IMP in humans, dogs, rabbits and rats, with two 99mTc labeled monoclonal antibodies in dogs and with 99mTc DTPA in renal transplants, we recommend this method as an adjunct in radiopharmaceutical development and evaluation.

Automated processing of first-pass radionuclide angiocardiography by factor analysis of dynamic structures.

A method for automatic processing of cardiac first-pass radionuclide study is presented. This technique, factor analysis of dynamic structures (FADS) provides an automatic separation of anatomical structures according to their different temporal behaviour, even if they are superimposed. FADS has been applied to 76 studies. A description of factor patterns obtained in various pathological categories is presented. FADS provides easy diagnosis of shunts and tricuspid insufficiency. Quantitative information derived from the factors (cardiac output and mean transit time) were compared to those obtained by the region of interest method. Using FADS, a higher correlation with cardiac catheterization was found for cardiac output calculation. Thus compared to the ROI method, FADS presents obvious advantages: a good separation of overlapping cardiac chambers is obtained; this operator independant method provides more objective and reproducible results. A number of parameters of the cardio-pulmonary function can be assessed by first-pass radionuclide angiocardiography (RNA) [1,2]. Usually, they are calculated using time-activity curves (TAC) from regions of interest (ROI) drawn on the cardiac chambers and the lungs. This method has two main drawbacks: (1) the lack of inter and intra-observers reproducibility; (2) the problem of crosstalk which affects the evaluation of the cardio-pulmonary performance. The crosstalk on planar imaging is due to anatomical superimposition of the cardiac chambers and lungs. The activity measured in any ROI is the sum of the activity in several organs and 'decontamination' of the TAC cannot easily be performed using the ROI method [3]. Factor analysis of dynamic structures (FADS) [4,5] can solve the two problems mentioned above. It provides an automatic separation of anatomical structures according to their different temporal behaviour, even if they are superimposed. The resulting factors are estimates of the time evolution of the activity in each structure (underlying physiological components), and the associated factor images are estimates of the spatial distribution of each factor. The aim of this study was to assess the reliability of FADS in first pass RNA and compare the results to those obtained by the ROI method which is generally considered as the routine procedure.

An interactive software for 81mKr pulmonary ventilation gating.

An interactive software to gate scintigraphic list-mode data collected from ventilation studies using a radioisotope of an inert gas (81mKr) has been developed. It allows the detection of end expiration and end inspiration phases in a time activity curve. It involves six phases: filtering of the curve; preselection of local extrema; validation of the preselection; choice of the parameters for the reconstruction; control of the selected cycles and display of parameters such as frequency, expiration time, breath amplitude; reconstruction of images representing the distribution of ventilation during a mean respiratory cycle. This software was tested on 116 data sets without a failure. The interactive definition and control of the selected times made the algorithm useful regardless of the count rate and the frequency. It could be adapted to gate other periodic or semiperiodic physiological phenomena.

Factor analysis of 81mKr lung ventilation studies.

Factor analysis of dynamic structures (FADS) summarizes data depending on time and space in a few elementary components. Each of them associates a time-activity curve (factor) and the spatial distribution of the corresponding events. The aim was to evaluate the patterns, the number of components, and their possible link to physiology when FADS was applied to scintigraphic images representing a composite of a 81mKr ventilation cycle. In a study of 26 patients (10 normal, 16 pathologic), components were found that represent: (1) a rapid and steeply changing ventilation factor, corresponding mainly to bases in normal subjects and whole lung fields in patients, (2) a slower expiration and shorter inspiration ventilation factor distributed throughout both lung fields in all patients, (3) a constant activity curve, with an inspiratory activity peak distributed over both lung fields and the large airways, and (4) a factor including a phase shift with respect to the first; this was found significantly more often in patients with pathology.

Factor analysis in gated cardiac studies.

Factor analysis of dynamic structures (FADS) can automatically provide "physiological" factors related to anatomical structures that have different temporal behavior, even if these structures overlap; it also yields images corresponding to the factors' spatial distributions. In normal patients, two significant cardiac factors, corresponding to the atria and the ventricles, may be extracted. A third significant factor can be obtained when additional dynamic structures exist. However, the method does not provide an estimate of the background. It becomes part of the factors, but it does not modify their shapes. FADS has been applied to 45 gated cardiac studies. Results obtained by FADS were compared with those obtained from the amplitude and the phase of first-harmonic Fourier analysis (FA). The joint results were compared with the final diagnosis, established by real-time echocardiography and/or ventriculography. In normal patients, good agreement was obtained between the two approaches. On the whole set of patients, FADS was significantly better than FA (by sign test: p less than 5%).

A co-operative study on the clinical value of dynamic renal scanning with deconvolution analysis.

An international project was set up to study the clinical usefulness of intrarenal transit times derived from the renogram by deconvolution. A common data sheet, to collect clinical, biochemical, radiological and isotopic information, was completed by the centres. Five hundred and ninety-one patients were studied and the results analysed. The mean transit time (MTT) in normal kidneys was found to be 3.6 +/- 1.1 min. If the MTT is greater than 7.6 min, a kidney is likely to be obstructed. In vesico-ureteric reflux, the transit times are prolonged, but they are normal in infection, hypertension, parenchymal disease and minimally irradiated kidneys. In transplantation, when the kidney is normal, the transit times are shorter than in the natural kidney; in acute rejection, transit time are prolonged.

Early kinetics of thyroid trap in normal human patients and in thyroid diseases.

A six-compartment model is proposed for the study of early iodine kinetics (150 min) in the human thyroid gland including three compartments for extrathyroidal spaces. The results were compared with those obtained with an open two-compartment thyroid model. In six euthyroid subjects the values of the unidirectional thyroid clearance R41 = 86 +- 25 ml/min and of the irreversible clearance R65 = 22 +- 9 ml/min were consistent with those generally accepted. With TSH stimulation a clear increase in the thyroidal pump effectiveness was found without an increase in the irreversible clearance. In four untreated thyrotoxic patients a tenfold increase in the unidirectional thyroid clearance was found (p less than 0.01). The binding rate was much less increased and overlapped with the normal values; a large increase in the iodide compartment volumes was found. The unidirectional clearance remained elevated in seven patients after Carbimazole treatment and the binding rates were reduced. The six-compartment model demonstrated a rapid exit from the bound iodine compartment. In three of four goitrous patients studied, the thyroid unidirectional clearance was significantly increased ( p less than 0.001). The volumes of the iodide compartments were much higher than those observed in the other groups. The method was applied to analyse the kinetics of the different parts of the thyroid.