Comparison of machine learning methods with logistic regression analysis in creating predictive models for risk of critical in-hospital events in COVID-19 patients on hospital admission

BACKGROUND

Machine learning (ML) algorithms have been trained to early predict critical in-hospital events from COVID-19 using patient data at admission, but little is known on how their performance compares with each other and/or with statistical logistic regression (LR). This prospective multicentre cohort study compares the performance of a LR and five ML models on the contribution of influencing predictors and predictor-to-event relationships on prediction model´s performance.

METHODS

We used 25 baseline variables of 490 COVID-19 patients admitted to 8 hospitals in Germany (March-November 2020) to develop and validate (75/25 random-split) 3 linear (L1 and L2 penalty, elastic net [EN]) and 2 non-linear (support vector machine [SVM] with radial kernel, random forest [RF]) ML approaches for predicting critical events defined by intensive care unit transfer, invasive ventilation and/or death (composite end-point: 181 patients). Models were compared for performance (area-under-the-receiver-operating characteristic-curve [AUC], Brier score) and predictor importance (performance-loss metrics, partial-dependence profiles).

RESULTS

Models performed close with a small benefit for LR (utilizing restricted cubic splines for non-linearity) and RF (AUC means: 0.763-0.731 [RF-L1]); Brier scores: 0.184-0.197 [LR-L1]). Top ranked predictor variables (consistently highest importance: C-reactive protein) were largely identical across models, except creatinine, which exhibited marginal (L1, L2, EN, SVM) or high/non-linear effects (LR, RF) on events.

CONCLUSIONS

Although the LR and ML models analysed showed no strong differences in performance and the most influencing predictors for COVID-19-related event prediction, our results indicate a predictive benefit from taking account for non-linear predictor-to-event relationships and effects. Future efforts should focus on leveraging data-driven ML technologies from static towards dynamic modelling solutions that continuously learn and adapt to changes in data environments during the evolving pandemic.

TRIAL REGISTRATION NUMBER

NCT04659187.

Clinical outcome, risk assessment, and seasonal variation in hospitalized COVID-19 patients-Results from the CORONA Germany study

Background

After one year of the pandemic and hints of seasonal patterns, temporal variations of in-hospital mortality in COVID-19 are widely unknown. Additionally, heterogeneous data regarding clinical indicators predicting disease severity has been published. However, there is a need for a risk stratification model integrating the effects on disease severity and mortality to support clinical decision-making.

Methods

We conducted a multicenter, observational, prospective, epidemiological cohort study at 45 hospitals in Germany. Until 1 January 2021, all hospitalized SARS CoV-2 positive patients were included. A comprehensive data set was collected in a cohort of seven hospitals. The primary objective was disease severity and prediction of mild, severe, and fatal cases. Ancillary analyses included a temporal analysis of all hospitalized COVID-19 patients for the entire year 2020.

Findings

A total of 4704 COVID-19 patients were hospitalized with a mortality rate of 19% (890/4704). Rates of mortality, need for ventilation, pneumonia, and respiratory insufficiency showed temporal variations, whereas age had a strong influence on the course of mortality. In cohort conducting analyses, prognostic factors for fatal/severe disease were: age (odds ratio (OR) 1.704, CI:[1.221–2.377]), respiratory rate (OR 1.688, CI:[1.222–2.333]), lactate dehydrogenase (LDH) (OR 1.312, CI:[1.015–1.695]), C-reactive protein (CRP) (OR 2.132, CI:[1.533–2.965]), and creatinine values (OR 2.573, CI:[1.593–4.154].

Conclusions

Age, respiratory rate, LDH, CRP, and creatinine at baseline are associated with all cause death, and need for ventilation/ICU treatment in a nationwide series of COVID 19 hospitalized patients. Especially age plays an important prognostic role. In-hospital mortality showed temporal variation during the year 2020, influenced by age.

Trial registration number

NCT04659187.

[Ultrasound versus MRI in preventive examinations - a retrospective analysis of 833 patients]

PURPOSE

The benefit of ultrasound in comparison with full-body MRI during a medical checkup in preventive health care was examined with regard to the detection of cardiovascular risk factors, metabolic syndrome, malignant tumors and further relevant findings.

MATERIALS AND METHODS

833 consecutive patients (266 f/567 m, age: 19 - 93 y, mean age: 56.6 y) underwent both ultrasound (extracranial carotid arteries, thyroid, abdominal ultrasound and echocardiography) and whole-body MRI (whole-body MR angiography, head, thorax, abdomen and virtual colonoscopy). For ultrasound examinations, DEGUM level III devices were used (Siemens Acuson Antares, Siemens G60, Siemens, Erlangen). MRI examinations were performed using a 1.5 Tesla MRI device (Siemens Avanto, Siemens, Erlangen). All patients were reviewed retrospectively based on the written reports.

RESULTS

Ultrasound was much more sensitive in detecting early atherosclerotic changes than MRI angiography. In 33 % of the patients, manifestations of atherosclerosis were found. Thoracic (3) and abdominal aortic and mesenteric artery aneurysms (3) were diagnosed by both methods. Hepatic steatosis as an important risk factor of metabolic syndrome was only found by ultrasound in 20.4 % of our patients. Malignant tumors were rare in this population (1.4 %): all abdominal tumors except one renal oncocytoma were found using both methods. MRI and ultrasound were equally sensitive with respect to the detection of small liver foci. As expected, MRI was less sensitive than ultrasound in the diagnosis of thyroid nodes. For intracranial diagnoses, malignant intrathoracic findings and colonic polyps, ultrasound is not the method of choice.

CONCLUSION

For the detection of lifestyle-dependent diseases such as atherosclerosis and metabolic syndrome, ultrasound examination was more sensitive than MRI, and the same was true for the early detection of thyroid diseases. For the detection of malignant abdominal tumors, both methods were equally sensitive. Whole-body MRI can additionally detect pathological changes in the head, lungs and colon.

Multicenter, double-blind, randomized, intraindividual crossover comparison of gadobenate dimeglumine and gadopentetate dimeglumine for MR angiography of peripheral arteries

PURPOSE

To prospectively compare the image quality and diagnostic performance achieved with doses of gadobenate dimeglumine and gadopentetate dimeglumine of 0.1 mmol per kilogram of body weight in patients undergoing contrast material-enhanced magnetic resonance (MR) angiography of the pelvis, thigh, and lower-leg (excluding foot) for suspected or known peripheral arterial occlusive disease.

MATERIALS AND METHODS

Institutional review board approval was granted from each center and informed written consent was obtained from all patients. Between November 2006 and January 2008, 96 patients (62 men, 34 women; mean age, 63.7 years +/- 10.4 [standard deviation]; range, 39-86 years) underwent two identical examinations at 1.5 T by using three-dimensional spoiled gradient-echo sequences and randomized 0.1-mmol/kg doses of each agent. Images were evaluated on-site for technical adequacy and quality of vessel visualization and offsite by three independent blinded readers for anatomic delineation and detection/exclusion of pathologic features. Comparative diagnostic performance was determined in 31 patients who underwent digital subtraction angiography. Data were analyzed by using the Wilcoxon signed-rank, McNemar, and Wald tests. Interreader agreement was determined by using generalized kappa statistics. Differences in quantitative contrast enhancement were assessed and a safety evaluation was performed.

RESULTS

Ninety-two patients received both agents. Significantly better performance (P < .0001; all evaluations) with gadobenate dimeglumine was noted on-site for technical adequacy and vessel visualization quality and offsite for anatomic delineation and detection/exclusion of pathologic features. Contrast enhancement (P < or = .0001) and detection of clinically relevant disease (P < or = .0028) were significantly improved with gadobenate dimeglumine. Interreader agreement for stenosis detection and grading was good to excellent (kappa = 0.749 and 0.805, respectively). Mild adverse events were reported for four (six events) and five (eight events) patients after gadobenate dimeglumine and gadopentetate dimeglumine, respectively.

CONCLUSION

Higher-quality vessel visualization, greater contrast enhancement, fewer technical failures, and improved diagnostic performance are obtained with gadobenate dimeglumine, relative to gadopentetate dimeglumine, when compared intraindividually at 0.1-mmol/kg doses in patients undergoing contrast-enhanced MR angiography for suspected peripheral arterial occlusive disease.

Combined arterial and venous whole-body MR angiography with cardiac MR imaging in patients with thromboembolic disease--initial experience

The objective was to assess the feasibility of a combined arterial and venous whole-body three-dimensional magnetic resonance (MR) angiography, together with a cardiac MR examination, in patients with arterial thromboembolism. Ten patients with arterial thromboembolism underwent a contrast-enhanced whole-body MR examination of the arterial and venous vessels, followed by a cardiac MR examination on a separate occasion within 24 h. All examinations were performed on a 1.5-T MR scanner. For both arterial and venous MR angiography only one injection of contrast agent was necessary. The cardiac imaging protocol included dark-blood-prepared half-Fourier acquisition single-shot turbo-spin-echo sequences, fast steady-state free precession cine sequences, T2-weighted turbo-spin-echo sequences and inversion recovery gradient-echo fast low-angle-shot sequences after injection of contrast agent. MR imaging revealed additional clinically unknown arterial thromboembolisms in four patients. The thoracic aorta was depicted as embolic source in four patients, while deep vein thrombosis (DVT) was found in one patient as the underlying disease. Unsuspected infarction of parenchymal organs was detected by MRI in two patients. An unknown additional DVT was found in one patient. Four patients were considered to have arterial emboli of cardiac origin. In conclusion, acquisition of arterial and venous MR angiograms of the entire vascular system combined with cardiac MR imaging is a most comprehensive and valuable strategy in patients with arterial thromboembolism.

[Diagnostic performance of contrast-enhanced MR angiography of the aortoiliac arteries with the blood pool agent Vasovist: initial results in comparison to intra-arterial DSA]

PURPOSE

To prospectively assess the accuracy of contrast-enhanced MR angiography of the aortoiliac arteries using the blood pool agent Vasovist compared to unenhanced time-of-flight MRA. Conventional digital subtraction angiography served as the standard of reference.

MATERIALS AND METHODS

Twenty-nine patients with suspected or known peripheral arterial occlusive disease (PAOD) were examined by means of contrast-enhanced aortoiliac MR angiography using a dosage of 0.03 mmol/kg bodyweight Vasovist. Unenhanced two-dimensional time-of-flight (TOF) MRA of the same anatomic region was performed immediately prior to injection of the contrast agent. This study was approved by the local Institutional Review Board and informed consent was obtained from all subjects. Both contrast-enhanced and unenhanced MRA images were compared to conventional angiography with respect to the presence of vascular stenosis. Three independent, blinded readers evaluated vessel stenosis and occlusion on the basis of DSA and MR angiographic image readings. Sensitivity, specificity, accuracy, and the area under the receiver operating characteristic curve were analyzed. Correlations between readers of conventional angiograms were calculated and compared to the MR results.

RESULTS

In the case of pooled readings, unenhanced TOF MR angiography had a sensitivity of 42.6 %, a specificity of 78.4 % and an accuracy of 74.9 % for detection of clinically significant (>50 %) stenosis. Vasovist-enhanced MRA showed significant improvement in sensitivity (83.3 %), specificity (88.8 %) and accuracy (88.3 %) compared to TOF MRA (p<0.01). The areas under the receiver operating characteristic curve for quantitative measurements increased significantly (p<0.01) in the case of Vasovist-enhanced MRA compared to TOF MRA. All readers found fewer images uninterpretable with Vasovist enhancement and the agreement regarding stenosis location and degree of stenosis between MR angiography and DSA improved substantially after Vasovist administration compared to the noncontrast examination.

CONCLUSION

MR angiography using the blood pool agent Vasovist is a feasible and minimally invasive alternative to DSA and provides angiograms of the aortoiliac region with high sensitivity, specificity, and diagnostic accuracy.

Contrast-Enhanced Magnetic Resonance Angiography

PURPOSE

To assess the gadolinium-based macromolecular intravascular contrast agent P792 for magnetic resonance angiography (MRA) at magnetic field strengths of 3.0 T, in comparison to 1.5 T, in rabbits.

MATERIALS AND METHODS

Eleven female New Zealand rabbits of the same age served as the animal model. Dose relationship testing was performed with 2 doses (13 and 25 micromol/kg; n = 4 per group) of P792 as compared with a single dose (100 micromol/kg; n = 3) of gadoterate meglumine (Gd-DOTA). All animals underwent contrast-enhanced MRA of the abdominal aorta and its branches on 2 occasions separated by 72 hours. The particular doses were administered in random order. Contrast-enhanced MRA was performed on 3.0 and 1.5 T whole-body MR systems, using a fast 3D spoiled gradient recalled echo sequence. Data acquisition was performed before and up to 10 minutes after administration of intravenous contrast material. Image quality was judged on a 4-point-Likert scale. Signal-to-noise and contrast-to-noise measurements were performed; statistical differences (P < 0.05) between the groups were determined.

RESULTS

P792 and Gd-DOTA yielded high-quality MR angiograms in rabbits in all cases. Although image quality within the first 3 minutes after contrast material administration was equal for both agents, P792 at a dose of 25 micromol/kg was considered superior to Gd-DOTA at the later time points. Signal-to-noise and contrast-to-noise values of the higher dose of P792 were statistically significantly higher than those of Gd-DOTA in the post-bolus phase.

CONCLUSIONS

P792 seems to be well suited for high-quality early phase and equilibrium phase MRA in rabbits at a field strength of 3.0 T, on the basis of this initial evaluation in an animal model.

Peripheral Vascular Disease: Comparison of Continuous MR Angiography and Conventional MR Angiography—Pilot Study

The aim of this study was to prospectively assess the accuracy of three-dimensional magnetic resonance (MR) angiography for evaluation of stenosis in the peripheral arterial system with a continuous moving table technique, with conventional MR angiography as reference. This study was approved by the local institutional review board; informed consent was obtained. Five healthy male volunteers (mean age, 27 years; range, 24-35 years) and four men and one woman (mean age, 63 years; range, 46-78 years) with peripheral arterial occlusive disease were examined. Images obtained with both techniques showed excellent concordance (Cohen kappa = 0.75). Images obtained with a conventional protocol had higher quality compared with those obtained with the continuous technique (mean, 1.07 +/- 0.25 [standard deviation] vs 1.58 +/- 0.6; P < .05); small vessels appeared sharper on them. For detection of significant stenosis and occlusion, accuracy, sensitivity, and specificity of the continuous technique were 92.8%, 100%, and 89.2%, respectively.

Clinical safety and diagnostic value of the gadolinium chelate gadoterate meglumine (Gd-DOTA)

OBJECTIVE

The purpose of this study was to assess the diagnostic value and safety of the contrast agent gadoterate meglumine (Gd-DOTA, DOTAREM, Guerbet, Roissy CdG Cedex, France) in the setting of a postmarketing surveillance study.

MATERIALS AND METHODS

Between January 2004 and October 2005, radiologists in 61 radiologic institutions were asked to document the routine use of Gd-DOTA in a questionnaire. In addition to assessing the image quality and diagnostic value of the contrast-enhanced magnetic resonance imaging scans, we statistically evaluated and analyzed demographic and safety data.

RESULTS

A total of 24,308 patients were intravenously injected with Gd-DOTA for various diagnostic examinations. The examination allowed for establishing a diagnosis in >99% of cases and image quality was rated as "excellent" or "good" in 97.5% of all cases. Adverse events were noted in only 0.4% of the examinations and were mostly rated as minor, such as feeling of warmth or taste alteration. There was one serious adverse event, albeit with complete recovery.

CONCLUSION

This postmarketing surveillance study suggests diagnostic efficacy and a favorable clinical safety profile of Gd-DOTA in clinical practice.

[Simultaneous manifestation of blue-rubber-bleb-nevus-syndrome and malignant melanoma]

Blue rubber bleb nevi (BRBN) are rare vascular malformations of the integument and mucous membranes. Little is known about coherence between these nevi and malignant diseases. We report on a young man with progressive BRBN syndrome representing a thitherto unknown malignant melanoma.

Osteosynthese im Gesichtsschädelbereich

BACKGROUND

The favorable properties of silicon nitride (Si3N4) ceramic, such as high stability and biocompatibility suggest its biomedical use as an implant material. The aim of this study was to test its suitability for osteosynthesis.

MATERIALS AND METHODS

A Si3N4 prototype minifixation system was manufactured and implanted for osteosynthesis of artificial frontal bone defects in three minipigs. After 3 months, histological sections, CT and MRI scans were obtained. Finite element modeling (FEM) was used to simulate stresses and strains on Si3N4 miniplates and screws to calculate survival probabilities.

RESULTS

Si3N4 miniplates and screws showed satisfactory intraoperative workability. There was no implant loss, displacement or fracture. Bone healing was complete in all animals and formation of new bone was observed in direct contact to the implants.

CONCLUSIONS

Si3N4 ceramic showed a good biocompatibility outcome both in vitro and in vivo. This ceramic may serve as biomaterial for osteosynthesis, e.g. of the midface including reconstruction of the floor of the orbit and the skull base. Advantages compared to titanium are no risk of implantation to bone with mucosal attachment, no need for explantation, no interference with radiological imaging.

Time-resolved contrast-enhanced magnetic resonance angiography of the hand with parallel imaging and view sharing: initial experience

We sought to compare a three-dimensional, contrast-enhanced, magnetic resonance angiogram (3D CE MRA) sequence combining parallel-imaging (generalised autocalibrating partially parallel acquisitions (GRAPPA)) with a time-resolved echo-shared angiographic technique (TREAT) in an intraindividual comparison to a standard 3D MRA sequence. Four healthy volunteers (27-32 years), and 11 patients (11-82 years) with vascular pathologies of the hand were examined on a 1.5-Tesla (T) MR system (Magnetom Avanto, Siemens, Erlangen, Germany) using two multichannel receiver coils. Following automatic injection (flow rate 2.5 cc/s) of 0.1 mmol/kg gadoterate (Dotarem, Guerbet, Roissy, France), 32 consecutive 3D data sets were collected with the TREAT sequence (TR/TE: 4.02/1.31 ms, FA: 10 degrees, GRAPPA acceleration factor: R=2, TREAT factor: 5, voxel size: 1.0 x 0.7 x 1.3 mm(3)) and a T1-wwighted 3D gradient-echo sequence (TR/TE: 5.3/1.57 ms, FA: 30 degrees, GRAPPA acceleration factor: 2, voxel size: 0.71 x 0.71 x 0.71 mm(3,)). MR data sets were evaluated and compared for image quality and visualisation of vascular details. In the volunteer group, all MR imaging was successful while technical problems prevented acquisition of the standard protocol in two patients. For the corresponding segments, the number of visible segments was equal on both sequences. Overall image quality was significantly better on the standard protocol than on the TREAT protocol. TREAT MRA provided functional information in lesions with rapid blood flow, e.g. detection of feeding and draining vessels in an haemangioma. TREAT-MRA is a robust technique that combines morphological and functional information of the hand vasculature and deals with the very special physiological demands of vascular lesions, such as quick arteriovenous transit time.

Renal Arteries: Comparison of Steady-State Free Precession MR Angiography and Contrast-enhanced MR Angiography

All participants provided informed consent to participate in this study, which was approved by the institutional review board. Breath-hold three-dimensional (3D) steady-state free precession (SSFP) magnetic resonance (MR) angiography was compared with 3D contrast material-enhanced MR angiography in patients suspected of having renal artery stenosis. Two radiologists assessed visualization of renal arteries and detection of vascular disease. With SSFP MR angiography, 39 of 41 renal arteries in 19 patients were correctly detected. Relevant stenoses were correctly identified with SSFP MR angiography in two patients. In two patients, SSFP MR angiographic data sets led to false-positive overgrading of vascular disease. Fast breath-hold 3D SSFP MR angiography appears to be feasible for MR angiography of renal arteries.

The Efficacy of Gadobenate Dimeglumine (Gd-BOPTA) at 3 Tesla in Brain Magnetic Resonance Imaging

OBJECTIVES

The objectives of this study were to analyze the differences in contrast enhancement using gadobenate dimeglumine (Gd-BOPTA or MultiHance) at 3 T versus 1.5 T and to compare Gd-BOPTA with a standard gadolinium chelate, gadopentetate dimeglumine (Gd-DTPA or Magnevist), at 3 T in a rat glioma model.

MATERIALS AND METHODS

Twelve rats with surgically implanted gliomas were randomized to either comparing Gd-BOPTA at 1.5 T versus 3 T (n=7) or comparing Gd-BOPTA and Gd-DTPA at 3 T (n=5). Matched T1-weighted spin-echo techniques were used for both comparisons and the order of examinations was randomized. Signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and lesion enhancement (LE) were evaluated using a region-of-interest analysis. A veterinary histopathologist evaluated all brain specimens.

RESULTS

In the evaluation of Gd-BOPTA at 3 T and 1.5 T, there were significant increases in SNR, LE, and CNR at 3 T. Average increases in brain and tumor SNR were 93% (P<0.0001) and 92% (P<0.0001), respectively. CNR increased by 121% (P<0.0001). Comparison of Gd-BOPTA and Gd-DTPA at 3 T demonstrated significantly higher CNR and LE with Gd-BOPTA. CNR increased by 35% (P=0.002). LE increased by 44% (P=0.03).

CONCLUSIONS

Gd-BOPTA provides significantly higher CNR at 3 T compared with 1.5 T and also demonstrates significantly higher CNR when compared with a standard Gd-chelate at 3 T. As a result of transient protein binding, Gd-BOPTA may be superior to standard gadolinium chelates in neurologic imaging at 3 T.

Magnetic resonance imaging of experimental atherosclerotic plaque: Comparison of two ultrasmall superparamagnetic particles of iron oxide

PURPOSE

To evaluate a new ultrasmall superparamagnetic particles of iron oxide (USPIO) compound, ferumoxytol, as a marker of macrophage activity in atherosclerotic plaques and to compare it to ferumoxtran-10.

MATERIALS AND METHODS

Ten mature heritable hyperlipidemic (WHHL) female Watanabe rabbits served as the animal model for atherosclerosis, four coeval female New Zealand White (NZW) rabbits were the control group. Five WHHL and two NZW received a single intravenous injection (250 micromol/kg) of either ferumoxtran-10 or ferumoxytol and were subjected to daily MR examinations on a 1.5T whole body scanner for the next five days. Development of signal intensity changes and susceptibility effects was assessed. Statistical analysis was based on a nonparametric Wilcoxon-Mann-Whitney-U test by using a P value at the 0.05 significance level. On day 5, the rabbits were sacrificed and the aorta was referred to histopathology, distribution of iron particles in the vessel wall was analyzed.

RESULTS

MRI was feasible in all animals. Three days after injection of ferumoxytol the highest luminal signal intensity measurements were observed in the ferumoxytol group; the highest measurements were five days after injection in the ferumoxtran-10 group (P < 0.05). In the WHHL, susceptibility effects presented as homogeneous dark lines parallel to the aortic wall after ferumoxytol and spotted areas void of signal after ferumoxtran-10. None of these findings were observed in the NZW control groups.

CONCLUSION

Ferumoxtran-10 and ferumoxytol at a respective dose of 250 mumol/kg appear well suited for atherosclerotic plaque detection with MRI in experimental atherosclerosis. Ferumoxytol warrants further analysis in humans.

Whole body positron emission tomography/computed tomography (PET/CT) tumour staging with integrated PET/CT colonography: technical feasibility and first experiences in patients with colorectal cancer

AIM

The aim of this study was to implement an imaging protocol for positron emission tomography/computed tomography (PET/CT) colonography and to combine this protocol with whole body PET/CT tumour staging for a single whole body examination for routine clinical use.

SUBJECTS AND METHODS

A whole body PET/CT protocol for tumour staging and a protocol for PET/CT colonography were integrated into one examination. Fourteen prospective patients with suspected colorectal cancer underwent whole body PET/CT after aqueous bowel distension and pharmacological bowel relaxation. Colonoscopy and histopathology served as the standards of reference in all patients.

RESULTS

The modified PET/CT examination detected all but one lesion in the colon. One additional lesion was detected in a patient with incomplete colonoscopy due to high grade luminal stenosis. One polyp with malignant conversion was identified with the modified PET/CT protocol. PET/CT colonography proved accurate in local lymph node staging and staged nine out of 11 patients correctly. Six additional extracolonic tumour sites were detected based on the whole body staging approach.

CONCLUSION

Whole body PET/CT with integrated colonography is technically feasible for whole body staging in patients with colorectal cancer. Based on these initial diagnostic experiences, this integrated protocol may be of substantial benefit in staging patients with colorectal cancer, focusing on patients with incomplete colonoscopy and those with small synchronous bowel lesions.

Contrast-enhanced dark lumen PET/CT and MR colonography in a rodent polyp model: initial results with histopathologic correlation

OBJECTIVE

The objective of this study was to assess the feasibility of PET/CT for the detection of colorectal masses in a rodent polyp model in an intraindividual comparison with dark-lumen MR colonography.

CONCLUSION

Detection of small tumors with PET/CT and MR colonography is possible in a rodent model. The technique thus warrants further evaluation in animal studies as well as in patients with suspected colorectal disease.

Whole-body MRI for staging patients with head and neck squamous cell carcinoma

CONCLUSIONS

Whole-body MRI is feasible for the tumor staging of patients with malignant head and neck tumors and appears to be a quick, reliable and proven alternative in general and for patients with contraindications to CT. This examination minimizes the logistical effort required compared to multimodality strategies. Its economic impact remains to be determined.

OBJECTIVE

To assess the performance of whole-body MRI for staging patients with squamous cell carcinoma of the head and neck region.

MATERIAL AND METHODS

This was a randomized, prospective clinical study. For tumor staging, 21 patients (mean age 56.7 years; range 43-80 years) with advanced malignant head and neck tumors underwent whole-body MRI in addition to routinely performed imaging investigations, including sonography, chest X-ray, CT of the head, neck and thorax and endoscopy. All investigations were accomplished within a period of 10+/-3 days in a random order. A randomized, blinded, consensus assessment of all the whole-body MRI examinations was performed by two radiologists. The localization and extent of the primary tumor and metastases were documented for whole-body MRI and compared to the standard of reference (all other imaging modalities as well as histology). Point estimates of the diagnostic accuracy of whole-body MRI were calculated.

RESULTS

In accordance with the standard of reference, the overall TNM category was correctly determined with whole-body MRI in all 21 patients. However, four patients were classified as having carcinoma of unknown primary, as the primary tumor was not found with any imaging modality. Two patients had mediastinal, pulmonary and hepatic metastases.

Total-body 3D magnetic resonance angiography influences the management of patients with peripheral arterial occlusive disease

High-resolution total-body 3D MR angiography (MRA) has recently become available, revealing additional clinically relevant disease in patients with peripheral arterial occlusive disease (PAOD). However, the actual impact of total-body MRA on patient management in patients with PAOD has not been investigated so far. Two hundred forty-nine consecutive patients with angiographically proven PAOD were prospectively examined by means of contrast-enhanced total-body 3D MRA on a 1.5-T MR scanner. All correlative imaging studies performed within 60 days of total-body MRA were included in the efficacy analysis. Additional clinically relevant disease (luminal narrowing >50%, aneurysmal changes or dissections) was found in 73 segments (52 patients), including the renal arteries (36 segments), carotid arteries (28 segments), subclavian arteries (four segments) and abdominal aortic aneurysms (AAA) (five segments). Of the 73 segments, 36 were deemed necessary for further investigation by means of focused MRA examinations; the diagnosis was confirmed in all cases. Within the 60-day follow-up period, interventional or surgical therapy outside the peripheral arterial tree was performed in nine patients (11 segments), including carotid endatherectomy and renal artery angioplasty. The outlined total-body 3D MRA approach permits a comprehensive evaluation of the arterial system in patients with atherosclerosis and does indeed have an impact on patient management in patients with PAOD.

Dark-lumen magnetic resonance colonography in patients with suspected sigmoid diverticulitis: a feasibility study

To assess dark-lumen magnetic resonance colonography (MRC) for the evaluation of patients with suspected sigmoid diverticulitis. Forty patients with suspected sigmoid diverticulitis underwent MRC within 72 h prior to conventional colonoscopy (CC). A three-dimensional T1-weighted volumetric interpolated breath-hold examination sequence was acquired after an aqueous enema and intravenous administration of gadolinium-based contrast agents. All MRC data were evaluated by two radiologists. Based on wall thickness and focal uptake of contrast material and pericolic reaction including mesenteric infiltration on T1-weighted sequence the sigmoid colon was assessed for the presence of diverticulitis. MRC classified 17 of the 40 patients as normal with regard to sigmoid diverticulitis. However, CC confirmed the presence of light inflammatory signs in four patients which were missed in MRC. MRC correctly identified wall thickness and contrast uptake of the sigmoid colon in the other 23 patients. In three of these patients false-positive findings were observed, and MRC classified the inflammation of the sigmoid colon as diverticulitis whereas CC and histopathology confirmed invasive carcinoma. MRC detected additionally relevant pathologies of the entire colon and could be performed in cases where CC was incomplete. MRC may be considered a promising alternative to CC for the detection of sigmoid diverticulitis.

Myocardial infarct: depiction with contrast-enhanced MR imaging--comparison of gadopentetate and gadobenate

Institutional review board approval and patient written informed consent were obtained. On two separate occasions, 24 hours apart, contrast-enhanced cardiac magnetic resonance (MR) imaging was performed prospectively at 1, 3, 5, 10, and 20 minutes after injection of gadopentetate dimeglumine and gadobenate dimeglumine in 15 patients (11 men, four women) with history of myocardial infarction. Both agents allowed detection of infarcted myocardium. T1 values at all times were significantly (P < .05) lower for gadobenate, compared with values for gadopentetate, in both infarcted and noninfarcted myocardium. At 1 minute after administration of both agents, T1 values in left ventricular cavity (LVC) were not different; at 3-20 minutes after injection, values were significantly (P < .05) lower for gadobenate. Differences between contrast-to-noise ratio (CNR) values of infarcted and noninfarcted myocardium were significantly higher on gadobenate-enhanced images (P < .05). CNR values between infarcted myocardium and LVC were significantly higher on gadopentetate-enhanced images (P < .05). Gadopentetate might permit better delineation of infarcts, especially subendocardial infarcts.

Magnetic resonance colonography without bowel cleansing using oral and rectal stool softeners (fecal cracking)—a feasibility study

The aim of our study was to assess the effect of oral and rectal stool softeners on dark-lumen magnetic resonance (MR) colonography without bowel cleansing. Ten volunteers underwent MR colonography without colonic cleansing. A baseline examination was performed without oral or rectal administration of stool softeners. In a second set, volunteers ingested 60 ml of lactulose 24 h prior to MR examination. In a third examination, water as a rectal enema was replaced by a solution of 0.5%-docusate sodium (DS). A fourth MR examination was performed, in conjunction with both oral administration of lactulose and rectal application of DS. A T1-weighted data set was acquired at scanning times of 0, 5 and 10 min after colonic filling. A fourth data set was acquired 75 s after i.v. injection of contrast agent. Signal intensity of stool was calculated for all colonic segments. Without oral ingestion of lactulose or rectal enema with DS stool signal intensity was high and did not decrease over time. However, lactulose and DS caused a decrease in stool signal intensity. Both substances together led to a decreasing signal intensity of feces. Combination of lactulose and DS provided the lowest signal intensity of stool. Thus, feces could hardly be distinguished from dark rectal enema allowing for the assessment of the colonic wall.

Evaluation of steady state free precession imaging of the pancreas

The aim of this work was to assess the diagnostic value of fast steady state free precession (SSFP) for the detection, characterization, and delineation of pancreatic lesions. Forty-eight patients referred for magnetic resonance (MR) imaging of the pancreas were included in the study. In addition to the standard protocol, axial pre-contrast SSFP slices of the pancreas were acquired. The standard of reference was defined as based on all imaging data other than SSFP, histopathology, surgery, and/or clinical follow-up. A randomized consensus reading of the SSFP data sets was retrospectively conducted by two board-certified radiologists. The presence of pancreatic lesions, local infiltration, and lymph node metastases was evaluated. Sensitivity and specificity were calculated and a receiver operating characteristic (ROC) analysis was performed. The overall sensitivity and specificity of SSFP were 0.93 and 0.77, respectively. Comparable values were achieved for lymph node detection (0.88/0.91) and assessment of vascular infiltration (0.94/0.91). The mean area under the ROC curve (Az) was 0.91. Owing to its potential to detect vascular infiltration and the rapid acquisition time, SSFP imaging should be supplemented as part of a standard MR protocol of the pancreas.

Assessment of Left Ventricular Parameters Using 16-MDCT and New Software for Endocardial and Epicardial Border Delineation

OBJECTIVE

The purpose of our study was to quantify left ventricular function and mass derived from retrospectively ECG-gated 16-MDCT coronary angiography data sets using a new analysis software based on automatic contour detection in comparison to corresponding standard of reference measurements acquired with MRI.

SUBJECTS AND METHODS

Multiplanar reformations in the short-axis orientation were calculated from axial contrast-enhanced CT images in 18 patients (men, 15; women, three; age range, 38-70 years; mean, 57.4 +/- 10.2 [SD] years) who were referred for CT coronary angiography. End-diastolic volume (EDV), end-systolic volume (ESV), ejection fraction (EF), and left ventricular mass (LVM) were analyzed with a recently developed imaging software using an automated contour detection algorithm of left ventricular endo- and epicardial contours and by manual tracing. The data were compared with similar measurements on MRI as the standard of reference.

RESULTS

EDV, ESV, EF, and LVM derived from an automated contour detection algorithm were not statistically significantly different from manual tracing (CT(auto) vs CT(manual): EDV = 137.1 +/- 45.7 mL vs 134.2 +/- 39.9 mL, ESV = 58.8 +/- 34.2 mL vs 58.1 +/-30.1 mL, EF = 59.2% +/- 13.7% vs 58.1% +/- 12.0%, LVM = 130.9 +/- 29.1 g vs 133.7 +/- 33.2 g; p > 0.05). However, EDV (118.7 +/- 43.6 mL), ESV (50.1 +/- 33.5 mL), and LVM (142.8 +/-38.4 g) as calculated on MR data sets were statistically significantly different from those calculated on CT (p < 0.05), whereas MRI-based EF (59.9% +/- 14.4%) did not differ statistically significantly from those based on both CT algorithms (p > 0.05).

CONCLUSION

Automatic and manual analysis of data acquired during CT coronary angiography using a 16-MDCT scanner allows a reliable assessment of left ventricular ejection fraction and a rough estimation of left ventricular volumes and mass.

Full-Body Cardiovascular and Tumor MRI for Early Detection of Disease: Feasibility and Initial Experience in 298 Subjects

OBJECTIVE

High diagnostic accuracy, emerging whole-body concepts, and lack of side effects combine to render MRI a natural candidate for screening purposes. The aim of this study was to evaluate the technical feasibility of a comprehensive multiorgan-targeting MRI examination and determine the frequency of findings in subjects without a history of serious disease.

SUBJECTS AND METHODS

The study group was composed of 331 subjects. The MRI protocol (mean examination time, 63 min) encompassed the target organs: the brain, arterial system, heart, and colon. Diagnoses were deemed relevant if the physician had to inform the subject about the findings. Subjects with a history of serious illnesses were excluded from subsequent analysis (n=33). All analyses were performed for the resulting subgroup of 298 subjects (247 men, 51 women; mean age, 49.7 years).

RESULTS

All 298 examinations were diagnostic excluding eight MR colonography components in which remaining stool hampered reliable diagnosis. Follow-up or radiologic confirmation could be obtained in 75% of all cases with relevant findings (128/169); only one false-positive result was encountered. Of the study group, 21% exhibited signs of atherosclerotic disease. Two cerebral infarctions and one myocardial infarction, previously unknown, were encountered; 12% had peripheral vascular disease. Twelve colonic polyps and nine pulmonary lesions were correctly detected. Of all MRI examinations, 29% revealed relevant additional findings in nontargeted organs. Only one minor allergoid reaction was encountered.

CONCLUSION

The presented data point toward an increased use of MRI for screening in the future, but to date screening MRI should not be performed outside a research setting because the cost-benefit relation is unclear.

[Cardiovascular whole-body MRI: possibilities and limitations in prevention]

Cardiovascular disease is a major challenge to the healthcare with increasing prevalence in western societies. Hence, early detection of cardiovascular pathologies and preventative strategies will experience growing relevance in the future. Magnetic resonance imaging (MRI) nowadays allows a comprehensive analysis of the cardiovascular system. By combining separate examinations of brain, arterial vasculature, and heart the technique permits early detection of pathological changes with high diagnostic accuracy void of adverse events. Such a protocol has been proven feasible and technically robust and can be performed within 45 min. Inherent limitations are low spatial resolution of whole-body MR angiography and lack of functional stress testing of the heart. However, while being suitable as a fast and comprehensive imaging technique for cardiovascular screening purposes, medical consequences and socioeconomic relevance must further be elucidated.

Digital subtraction dark-lumen MR colonography: Initial experience

PURPOSE

To evaluate image subtraction for the detection of colonic pathologies in a dark-lumen MR colonography exam.

MATERIALS AND METHODS

A total of 20 patients (12 males; 8 females; mean 51.4 years of age) underwent MR colonography after standard cleansing and a rectal water enema on a 1.5-T whole-body MR system. After suppression of peristaltic motion, native and Gd-contrast-enhanced three-dimensional T1-w gradient echo images were acquired in the coronal plane. Two radiologists analyzed the MR data sets in consensus on two separate occasions, with and without the subtracted images for lesion detection, and assessed the value of the subtracted data set on a five-point Likert scale (1=very helpful to 5=very unhelpful). All imaging results were compared with endoscopy.

RESULTS

Without subtracted images, MR-colonography detected a total of five polyps, two inflammatory lesions, and one carcinoma in eight patients, which were all verified by endoscopy. Using subtraction, an additional polyp was found, and readout time was significantly shorter (6:41 vs. 7:39 minutes; P<0.05). In two patients, endoscopy detected a flat adenoma and a polyp (0.4 cm) that were missed in the MR exam. Sensitivity and specificity without subtraction were 0.67/1.0, and 0.76/1.0 with the subtracted images, respectively. Subtraction was assessed as helpful in all exams (mean value 1.8+/-0.5; Likert scale).

CONCLUSION

We consider subtraction of native from contrast-enhanced dark-lumen MR colonography data sets as a beneficial supplement to the exam.

Venous Compression at High-Spatial-Resolution Three-dimensional MR Angiography of Peripheral Arteries

The aim of this study was to assess a venous compression technique that is performed with contrast material-enhanced peripheral magnetic resonance (MR) angiography to reduce venous enhancement. Healthy volunteers, as well as patients with correlating digital subtraction angiographic (DSA) findings, were examined. Venous compression was accomplished by placing a cuff at the midfemoral level unilaterally. Arterial signal-to-noise and contrast-to-noise ratios indicated no significant differences between compressed and noncompressed legs. Venous overlay was substantially reduced in the compressed legs. MR angiography with venous compression yielded diagnostic image quality and results that had excellent correlation with DSA findings. High-spatial-resolution peripheral MR angiography of improved diagnostic quality appears feasible, even with long data acquisition times.

Dark Lumen Magnetic Resonance Colonography in a Rodent Polyp Model

PURPOSE

We sought to assess dark lumen magnetic resonance (MR) colonography for the detection of colon polyps in a rodent model with histology as the gold standard.

MATERIAL AND METHODS

Fourteen male Wistar rats were subjected to carcinogenic N-methyl-N'-nitro-N-nitrosoguanidine at the age of 4 months to induce colon neoplasms. MR imaging was performed after a time interval of 1 year. Preparation and data acquisition was performed with the animals under full anesthesia. After a body-warm saline enema images were acquired on a clinical 1.5-T whole-body MR system using a standard extremity coil. Plain and contrast-enhanced (0.3 mmol/kg; Gd-DOTA; Dotarem, Guerbet, France) 3-dimensional T1-weighted gradient recall echo images were acquired. Two radiologists analyzed the MR data sets in consensus for lesion depiction. Contrast uptake in colonic wall and polyps was quantitatively assessed by signal-to-noise ratio and contrast-to-noise ratio measurements and compared using a Wilcoxon-Mann-Whitney U test with statistical significance at a P value < 0.05. Finally, all animals were killed, and the MR imaging results were compared with pathologic findings. Sensitivity and specificity were calculated.

RESULTS

By pathology, a total of 15 polyps were found in 9 of 14 rats. MR colonography detected 13 of 15 polyps measuring between 4 and 11 mm (mean 7 +/- 0.6 mm) in 8 of 9 animals, resulting in a sensitivity and specificity of 0.87 and 1.0, respectively. Compared with the precontrast data, all polyps showed a statistically significant increase in signal-to-noise ratio (78.2 +/- 6.3 to 167.4 +/- 17.7) and contrast-to-noise ratio (45.4 +/- 5.2 to 124.6 +/- 11.2).

CONCLUSION

MR colonography with a dark colon lumen and a bright, contrast-enhanced colon wall appears well suited for the detection of colonic lesions in a rodent model.

MR Coronary Angiography with SH L 643 A: Initial Experience in Patients with Coronary Artery Disease

PURPOSE

To prospectively assess the accuracy of breath-hold three-dimensional magnetic resonance (MR) coronary angiography with the gadolinium-based intravascular contrast agent SH L 643 A in patients with coronary artery disease.

MATERIALS AND METHODS

Twelve patients (seven men, five women; age range, 46-78 years; mean age, 61.3 years) with angiographically proved coronary artery disease (luminal narrowing >50%) underwent breath-hold three-dimensional MR coronary angiography before and after injection of SH L 643 A (0.1 mmol gadolinium per kilogram body weight). For all MR examinations, signal-to-noise ratio and contrast-to-noise ratio were measured. Image quality was assessed with a four-point scale. Conventional angiograms and MR angiograms were evaluated for depiction of the left main, proximal and middle left anterior descending, proximal left circumflex, and proximal and middle right coronary artery segments in a blinded fashion by two experienced readers in consensus. Results of this evaluation were compared by using a paired Student t test. P < .05 was considered to indicate a statistically significant difference.

RESULTS

For the 72 coronary artery segments, the contrast-to-noise ratio significantly improved after administration of SH L 643 A, compared with the prior ratio (9.8 +/- 5.1 [standard deviation] vs 23.0 +/- 8.7; P < .01), whereas the difference in signal-to-noise ratio did not reach statistical significance (25.2 +/- 11.4 vs 29.5 +/- 9.8; P > .3). Image quality significantly improved from a mean of 2.0 +/- 0.9 for nonenhanced images to 2.9 +/- 0.9 (P < .03) for contrast material-enhanced images. The proportion of segments for which images were nondiagnostic decreased from 38% to 10% with application of SH L 643 A. Overall sensitivity and specificity of contrast-enhanced MR coronary angiography for detection of coronary artery disease were 80% and 93%, respectively, and accuracy was 87%.

CONCLUSION

Use of SH L 643 A improves detection of coronary artery disease at three-dimensional MR coronary angiography.

Whole-Body MR Imaging: Evaluation of Patients for Metastases

PURPOSE

To compare the results of whole-body magnetic resonance (MR) imaging with staging based on computed tomographic (CT), dedicated MR imaging, and nuclear scintigraphic results as standard of reference.

MATERIALS AND METHODS

Fifty-one patients with known malignant tumors were included in the study. Patients were placed on a rolling table platform capable of moving the patient rapidly through the isocenter of the magnet bore. The thorax and the abdomen were imaged by using fast breath-hold T2-weighted sequences in the transverse plane. After intravenous administration of a paramagnetic contrast agent, three-dimensional gradient-echo data sets were collected in five stations and covered the body from the skull to the knees. Location and size of cerebral, pulmonary, hepatic, and osseous metastases were documented by two experienced radiologists. Whole-body MR imaging findings were compared with results obtained at skeletal scintigraphy, CT, and dedicated MR imaging.

RESULTS

The mean examination time for whole-body MR imaging was 14.5 minutes. All cerebral, pulmonary, and hepatic metastases greater than 6 mm in diameter could be identified with whole-body MR imaging. Small pulmonary metastases were missed with MR imaging, which did not change therapeutic strategies, but MR imaging depicted a single hepatic metastasis that was missed with CT. Skeletal scintigraphy depicted osseous metastases in 21 patients, whereas whole-body MR imaging revealed osseous metastases in 24 patients. The additional osseous metastases seen with MR imaging were confirmed at follow-up examinations but did not result in a change in therapy. Whole-body MR imaging performed on a per-patient basis revealed sensitivity and specificity values of 100%.

CONCLUSION

Whole-body MR imaging for the evaluation of metastases compared well with the reference techniques for cerebral, pulmonary, and hepatic lesions. Whole-body MR imaging was more sensitive in the detection of hepatic and osseous metastases than were the reference techniques.

HASTE MRI Versus Chest Radiography in the Detection of Pulmonary Nodules: Comparison with MDCT

OBJECTIVE

The purpose of our study was to compare the diagnostic accuracy of an ultrafast ECG-triggered black blood-prepared HASTE sequence with chest radiography for the detection of pulmonary nodules. SUBJECTS AND METHODS. Sixty-four patients with various primary malignancies who had undergone radiography and MDCT of the chest also underwent ECG-triggered black blood-prepared HASTE MRI of the lung. MR images and radiographs were interpreted separately. The number, location, and size of detected lesions were recorded, and each hemithorax was classified as affected or not affected on the basis of a grade reflecting the conspicuity of nodular involvement. Sensitivity, specificity, and positive and negative predictive values for the detection of pulmonary nodules with diameters of 5 mm or larger were determined, using MDCT findings as the standard of reference. Lesions with diameters smaller than 5 mm were not evaluated. Additional lesion-by-lesion comparisons between MDCT and MRI findings were performed.

RESULTS

MDCT confirmed pulmonary lesions in 32 patients, whereas HASTE MRI revealed lesions in 30 patients and chest radiography, in 19 patients. MDCT revealed 226 nodules in 32 patients, whereas MRI HASTE revealed 227 lesions in 30 patients. Conspicuity scale-based sensitivity and specificity for chest radiography were 55.8% and 92.4%, respectively, whereas HASTE MRI had a sensitivity of 93.0% and a specificity of 96.2%. Positive and negative predictive values for chest radiography were 80% and 79.3%, respectively, and for HASTE MRI, 93.0% and 96.2%, respectively. The sensitivity of HASTE MRI increased with lesion size, ranging from 94.9% for nodules between 5 and 10 mm in diameter to 100% for lesions exceeding 3 cm in diameter.

CONCLUSION

ECG-triggered black blood-prepared HASTE MRI is reliable for detecting pulmonary nodules exceeding 5 mm and has proven significantly more accurate than conventional chest radiography. The technique appears useful as an adjunct to MRI of the heart, great vessels, or chest, potentially increasing the diagnostic yield of MRI examinations.

Whole-Body 3D MR Angiography of Patients with Peripheral Arterial Occlusive Disease

OBJECTIVE

We assessed the diagnostic performance of whole-body 3D contrast-enhanced MR angiography in comparison with digital subtraction angiography (DSA) of the lower extremities in patients with peripheral arterial occlusive disease. SUBJECTS AND METHODS. Fifty-one patients with clinically documented peripheral arterial occlusive disease referred for DSA of the lower extremity arterial system underwent whole-body MR angiography on a 1.5-T MR scanner. Paramagnetic gadobutrol was administered and five contiguous stations were acquired with 3D T1-weighted gradient-echo sequences in a total scanning time of 72 sec. DSA was available as a reference standard for the peripheral vasculature in all patients. Separate blinded data analyses were performed by two radiologists. Additional vascular disease detected by whole-body MR angiography was subsequently assessed on sonography, dedicated MR angiography, or both.

RESULTS

All whole-body MR angiography examinations were feasible and well tolerated. AngioSURF-based whole-body MR angiography had overall sensitivities of 92.3% and 93.1% (both 95% confidence intervals [CIs], 78-100%) with specificities of 89.2% and 87.6% (both CIs, 84-98%) and excellent interobserver agreement (kappa = 0.82) for the detection of high-grade stenoses. Additional vascular disease was detected in 12 patients (23%).

CONCLUSION

Whole-body MR angiography permits a rapid, noninvasive, and accurate evaluation of the lower peripheral arterial system in patients with peripheral arterial occlusive disease, and it may allow identification of additional relevant vascular disease that was previously undetected.