Biomedical subjects
Publications and source records attributed to D D Stark.
Magnetic resonance is potentially a second imaging modality that can be used to study the fetal cardiovascular system. Magnetic resonance imaging to date has not been associated with any harmful effects and, since it does not involve ionizing radiation, we are evaluating its role in fetal imaging. We report a case in which this modality showed the unique features of the cardiovascular system.
Magnetic resonance (MR) imaging distinguished hepatitis from fatty liver and cirrhosis in a woman with a history of alcohol abuse. Anatomic and physiologic manifestations of portal hypertension were also demonstrated by MR.
Nineteen patients with proven pancreatic disease and 50 control subjects were examined by magnetic resonance (MR) using a variety of spin echo and inversion recovery techniques. The MR results were then compared with CT scans. The normal pancreatic head, body, and tail were identified by MR in approximately 60% of patients. Pancreatic adenocarcinoma and retroperitoneal lymphoma were detected using morphologic criteria similar to those used in CT. Differentiating bowel from pancreas was difficult on MR in patients with little retroperitoneal fat, and tissue relaxation times were usually not helpful in differentiating adenocarcinoma or lymphoma from normal pancreatic tissue. However, MR intensity, T1, and T2 were useful in differentiating pancreatic islet cell tumors from normal pancreatic tissue. MR accurately identified retroperitoneal invasion, vascular involvement, and liver metastases. In pancreatitis, tissue T1 and T2 relaxation times were prolonged and complications such as ductal dilatation, pseudocyst, phlegmon, and ascites were identified. Small pancreatic calcifications were not detected by MR. Pancreatic iron overload was seen in patients with hemochromatosis. Although respiratory motion and spatial resolution are currently limiting factors, MR is a versatile and unique modality for the evaluation of pancreatic disease.
In 41 patients with clinical evidence of pancreatic islet cell tumors, CT detected 21 of 27 tumors that were proved at surgery (sensitivity 78%) and correctly excluded the diagnosis in 14 of 14 patients with no evidence of tumor at long-term follow-up. Using rapid sequence CT scanning and bolus contrast material infusion, a vascular blush was identified in 13 of 16 tumors (81%). Tumors as small as 0.9 cm were localized. CT identified all malignant tumors, and staging accuracy was 100%. CT is useful as the initial imaging procedure for the diagnosis and staging of suspected islet cell tumors. Pancreatic venous sampling and angiography are complementary to CT for the detection of smaller lesions.
Magnetic resonance (MR) images of the neck were obtained in 24 patients using a specially designed radio-frequency coil, standard and high-resolution imaging techniques, and a variety of spin echo and inversion recovery pulse sequences. Cervical vascular structures were more easily identified with MR than with CT because of the inherent contrast of flowing blood. The laryngeal skeleton, paralaryngeal tissues, trachea, thyroid, esophagus, and muscles were identified in all cases by MR using morphologic criteria comparable with CT criteria. The spatial resolution of high-resolution MR was slightly inferior to high-resolution CT. However, streak artifacts due to motion and x-ray beam hardening often limited CT, but did not affect MR. Anatomic resolution was best using high-resolution spin echo images obtained with TR = 2.0 seconds and TE = 28 msec because this imaging technique offered excellent contrast between normal tissues and had the highest signal-to-noise ratio. With further improvements in spatial resolution MR is likely to become an important technique for imaging the neck.
Magnetic resonance (MR) images of the neck were obtained in 14 patients with thyroid, parathyroid, lymph node, or laryngeal lesions. Tumors and lymph nodes were more easily differentiated from muscle and blood vessels with MR than with CT because of the superior soft tissue contrast of MR. Tissue characterization allowed MR differentiation of thyroid nodules, thyroid cysts, and parathyroid tumors from normal thyroid tissue; however, nonspecifically increased T1 and T2 relaxation times overlapped for a variety of neoplastic and inflammatory conditions. Thyroid cyst fluid had the greatest water content and longest T1 and T2 times of all tissues studied. Parathyroid hyperplasia could not be differentiated from parathyroid adenoma; however, parathyroid tumors had slightly longer T1 and T2 times than thyroid nodules or lymph nodes. With further experience, MR tissue characterization may become a useful technique for evaluating neck masses.
Twenty normal subjects and ten patients with pericardial abnormalities underwent ECG-gated magnetic resonance (MR) imaging of the thorax using a 0.35-tesla superconducting system. The patients with pericardial abnormalities were also evaluated with serial chest radiographs, ultrasound, computed tomography, and/or angiography. ECG gating was necessary to identify the normal pericardium, which was visualized as a 1- to 2-mm-wide curvilinear structure of low signal intensity. Pericardial thickening in constrictive pericarditis was clearly delineated on gated MR images. Pericardial inflammation caused a marked increase in signal intensity as well as thickening of the pericardium. Pericardial effusions and pericardial adhesions were also demonstrated. A simple pericardial cyst and a complex pericardial mass were identified and differentiated from pericardial fat and diaphragmatic eventration. MR appears to be an important modality for the evaluation of pericardial disease.
Eight patients who had benign mediastinal masses underwent CT and magnetic resonance (MR) imaging. The intensity difference between mass and surrounding mediastinal fat was greater on MR than CT images. Although the masses were clearly identified with both modalities, delineation of a mass from surrounding mediastinal structures was better with CT. Heterogeneity on CT and MR images was a prominent feature of most larger benign mediastinal masses, although this should not be considered indicative of benignity. These benign mediastinal masses were best demonstrated on MR images obtained using a long TR and short TE. In general, the MR studies provided information comparable to but not identical to that of CT scanning. The viewing of mediastinal masses using a variety of TR and TE imaging parameters appeared to be an important aspect of the MR studies.
Magnetic resonance (MR) images of the hila were reviewed in 25 normal subjects and 12 patients with unilateral or bilateral hilar masses. On spin echo MR images in normal patients, collections of soft tissue large enough to be confused with an abnormally enlarged lymph node were seen in three locations. In patients with a hilar mass, the mass was differentiated from hilar vasculature more easily using MR than contrast-enhanced CT. In five of these patients, hilar lymph nodes approximately 1 cm in diameter were easily seen using MR, but were difficult or impossible to appreciate prospectively on CT. However, because the spatial resolution of MR is inferior to that of CT, bronchi were difficult to evaluate using MR. In general, images obtained with a short TR (0.5 sec) provided the best definition of mediastinal extension of the hilar mass, and images with a longer TR (1.5 to 2.0 sec) provided a better signal-to-noise ratio and showed increased signal strength from hilar masses. Electrocardiographic-gated images showed better resolution of hilar structures but may not be necessary for large masses.
The response of different magnetic resonance (MR) techniques to tissue parameters - T1, T2, and N(H) - is a determinant of clinical efficacy. The large possible number of imaging techniques and range of variable parameters for each make it difficult to perform exhaustive evaluations in a single patient or even in animal models. In addition, changes in operating magnetic field strength change the relaxation times sufficiently so that the efficacy of a technique at a given field does not imply similar results at another field value. Tools are demonstrated that permit the evaluation of the efficacy of any model of response to tissue parameters, and also allow the investigation of the effects of changing magnetic field. Global maps of signal difference between tissues as well as calculated images are obtainable from a minimally sufficient data set. These tools serve as an important adjunct to experimental work.
The effect of periportal collagen deposition on magnetic resonance images and T1 and T2 relaxation times was studied in the rat. Hepatic cirrhosis was induced in 29 rats by chronic intraperitoneal thioacetamide injections. Another 14 rats in which liver abnormalities did not develop were used as controls. The rats were imaged using a small-bore resistive magnet. Histologic correlations and hydroxyproline measurements were performed to document the changes in periportal collagen deposition. The T1 and T2 relaxation times, determined both in vivo and in vitro with spectroscopy, were compared between the normal group and the group with moderate to severe histologic evidence of cirrhosis. The deposition of two to four times the normal amount of collagen in the liver did not affect the T1 or T2 relaxation time. Relatively pure periportal collagen fibrosis does not appear to affect the magnetic resonance image or T1 or T2 relaxation times of the rat liver.
Ten patients having a mediastinal tumor were studied with magnetic resonance imaging (MRI) using from two to four imaging sequences. Seven had bronchial carcinoma and three had benign lesions. The sequences included the spin-echo technique with repetition time (TR) values of 0.5, 1.0, and 2.0 sec and echo time (TE) values of 28 and 56 msec, and the inversion-recovery technique. The signal-intensity ratios of the mediastinal mass and mediastinal fat, which are a measure of image contrast, were compared for the different imaging sequences. Also signal-to-noise ratios were measured relative to both mediastinal fat and mediastinal mass. With spin-echo imaging, decreasing the TR value resulted in an increase in mass/fat contrast in all patients, making the masses easier to detect, but this also resulted in decreased signal-to-noise ratios. Inversion-recovery imaging with the sequence used resulted in a greatly increased mass/fat contrast, because of a relative decrease in signal from the mass. However, in two of four patients studied with this technique, the mass was so low in intensity that it could not be distinguished from the trachea or mediastinal vessels, and in one of these four, a lung nodule was also invisible. Spin-echo imaging with both short and long TR values provides good tissue contrast and good signal-to-noise ratios.
Magnetic resonance (MR) images of the neck were obtained in 16 patients with use of a variety of spin echo and inversion recovery pulse sequences. Anatomic resolution was best with high-resolution spin echo images obtained with the pulse sequence repetition rate equal to 2.0 seconds and the echo delay equal to 28 msec because this imaging technique offered excellent contrast between normal tissues and had the highest signal noise ratio. The spatial resolution of MR was nearly as good as state of the art computerized tomography (CT). However, streak artifacts caused by motion and x-ray beam hardening often limited CT but did not affect MR. Tumors and lymph nodes were more easily differentiated from muscle and blood vessels with MR than with CT because of the superior soft-tissue contrast of MR. Tissue characterization allowed MR differentiation of thyroid nodules, thyroid cysts, and parathyroid tumors from normal thyroid tissue. Thyroid cyst fluid had the greatest water content and longest T1 and T2 relaxation times of all tissues studied. However, nonspecifically increased T1 and T2 relaxation times overlapped for a variety of neoplastic and inflammatory conditions. With further experience, MR imaging is likely to become a useful technique for the evaluation of neck masses.
Results of computed tomography (CT), scintigraphy, excretory urography, and other imaging tests used to diagnose and stage 38 cases of neuroblastoma prior to treatment were reviewed. Findings of these examinations were correlated with clinical data, laboratory data, results of biopsy, and surgical findings. CT was the most sensitive single test (100%) for the detection and delineation of the primary tumor. Calcifications that suggested the histologic diagnosis of neuroblastoma were present in 79% of the cases. Rim calcifications, the most specific pattern for neuroblastoma, were identified in 29% of all cases. CT alone accurately staged 82% of cases; when complemented by bone marrow biopsy, staging accuracy was 97%. CT alone was more accurate than any combination of imaging tests that excluded CT. An algorithm using CT is presented for the diagnosis and staging of neuroblastoma at reduced cost and with increased efficiency.
One hundred twelve CT scans of 52 patients who were receiving treatment for neuroblastoma were reviewed for accuracy, and findings were correlated with data obtained from other imaging tests, physical examinations, laboratory tests, biopsies, surgery, and long-term clinical follow-up. CT was the most sensitive imaging test for tumor recurrence (85% detection rate), and it was also the most versatile in the ability to define recurrent disease in the retroperitoneum, liver, cranium, mediastinum, lymph nodes, and skeleton. All 30 tumor recurrences were detected by the combination of CT, bone-marrow biopsy, and selected spot radiographs at the sites of pain. CT was accurate and clinically useful both for assessing tumor response to therapy and for predicting findings at "second look" surgery (accuracy, 94%). CT is relatively cost effective by substituting for a more expensive but less accurate combination of competitive imaging studies.
Conventional parathyroid computed-tomographic scanning detected 51% (29/57) of tumors averaging 1.5 X 1.0 X 0.8 cm in size at surgery. Forty-six per cent (7/43) of the scans were uninterpretable due to shoulder-streak artifacts. Using a specially designed patient positioning device and a new imaging protocol involving dynamic scanning and the intravenous administration of a bolus of contrast material, 14 patients were scanned; 78% (18/23) of tumors averaging 1.4 X 1.0 X 0.8 cm in size were detected. All 14 scans were interpretable, and image quality was markedly improved.
Experimental animal models of hepatitis, fatty liver, and hepatic iron overload were evaluated using a 3.5-kGauss nuclear magnetic resonance (NMR) imaging system. Increases in image intensity measurements and in T2 relaxation times equalled the sensitivity of histologic findings for the detection of early stages of hepatitis. A significant shift in T1 relaxation times characterized the early stages of hepatic necrosis. Liver triglyceride content correlated significantly with increases in NMR intensity measurements (p less than 0.01); however, changes in liver water content had a much greater influence on intensity, T1, and T2. Thus, it may be possible to distinguish hepatitis from benign fatty liver. Liver iron content correlated with decreases in NMR intensity measurements (p less than 0.001), and iron levels as low as 1.2 mg/g were detected. NMR may more specifically identify hepatocellular iron overload than do other techniques that do not distinguish hepatocellular from reticuloendothelial iron.
Conventional chest radiographs and computed tomographic (CT) scans of 70 inflammatory thoracic lesions in 63 patients were reviewed and scored for diagnostic features. Pathologic confirmation of the final diagnosis was available in 42% (5/12) of lung abscesses and 31% (18/58) of empyemas. CT alone was sufficient to correctly diagnose 100% (70/70) of cases. Diagnostic information not available from conventional chest radiographs was obtained in 47% (33/70) of cases; in an additional 34% of patients, CT more accurately defined the extent of disease. The most reliable CT features for the differential diagnosis of lung abscess and empyema were wall characteristics, pleural separation, and lung compression. Conventional radiographic features such as size, shape, and the angle of the lesion with the chest wall were less helpful, though also best assessed by CT.