Transformer-based structuring of free-text radiology report databases.

OBJECTIVES: To provide insights for on-site development of transformer-based structuring of free-text report databases by investigating different labeling and pre-training strategies. METHODS: A total of 93,368 German chest X-ray reports from 20,912 intensive care unit (ICU) patients were included. Two labeling strategies were investigated to tag six findings of the attending radiologist. First, a system based on human-defined rules was applied for annotation of all reports (termed "silver labels"). Second, 18,000 reports were manually annotated in 197 h (termed "gold labels") of which 10% were used for testing. An on-site pre-trained model (Tmlm) using masked-language modeling (MLM) was compared to a public, medically pre-trained model (Tmed). Both models were fine-tuned on silver labels only, gold labels only, and first with silver and then gold labels (hybrid training) for text classification, using varying numbers (N: 500, 1000, 2000, 3500, 7000, 14,580) of gold labels. Macro-averaged F1-scores (MAF1) in percent were calculated with 95% confidence intervals (CI). RESULTS: Tmlm,gold (95.5 [94.5-96.3]) showed significantly higher MAF1 than Tmed,silver (75.0 [73.4-76.5]) and Tmlm,silver (75.2 [73.6-76.7]), but not significantly higher MAF1 than Tmed,gold (94.7 [93.6-95.6]), Tmed,hybrid (94.9 [93.9-95.8]), and Tmlm,hybrid (95.2 [94.3-96.0]). When using 7000 or less gold-labeled reports, Tmlm,gold (N: 7000, 94.7 [93.5-95.7]) showed significantly higher MAF1 than Tmed,gold (N: 7000, 91.5 [90.0-92.8]). With at least 2000 gold-labeled reports, utilizing silver labels did not lead to significant improvement of Tmlm,hybrid (N: 2000, 91.8 [90.4-93.2]) over Tmlm,gold (N: 2000, 91.4 [89.9-92.8]). CONCLUSIONS: Custom pre-training of transformers and fine-tuning on manual annotations promises to be an efficient strategy to unlock report databases for data-driven medicine. KEY POINTS: • On-site development of natural language processing methods that retrospectively unlock free-text databases of radiology clinics for data-driven medicine is of great interest. • For clinics seeking to develop methods on-site for retrospective structuring of a report database of a certain department, it remains unclear which of previously proposed strategies for labeling reports and pre-training models is the most appropriate in context of, e.g., available annotator time. • Using a custom pre-trained transformer model, along with a little annotation effort, promises to be an efficient way to retrospectively structure radiological databases, even if not millions of reports are available for pre-training.

[Classification of pancreatitis]. / Klassifikation der Pankreatitis.

BACKGROUND: Both, acute and chronic pancreatitis represent complex disease patterns, whose effective treatment is based on structured diagnostics and therapy by a multi-professional team. There are different systems for an improved objectivity in the classification of the severity of the disease OBJECTIVES: Overview of the common classifications of acute and chronic pancreatitis. MATERIALS AND METHODS: Literature research of currently used classifications of acute and chronic pancreatitis. Evaluation of the current chronic pancreatitis guideline. RESULTS: For acute pancreatitis, the modified Atlanta Classification and the "determinant-based" classification are most widely used. These classifications are complemented by clinical risk scores such as the Marshall Score and the SOFA Score. In chronic pancreatitis, the image-based Cambridge classification was established. For clinical assessment further classifications such as the ABC classification and the M­ANNHEIM classification are applied evaluating leading symptoms such as pain, exocrine and endocrine pancreatic insufficiency.

[CT and MRI of the liver: when, what, why?] / CT und MRT der Leber: wann, was, warum?

The detection and differential diagnostic clarification of liver pathologies play an important role in almost all medical disciplines. Because of its superior soft tissue contrast, the availability of liver-specific contrast agents and functional techniques, magnetic resonance imaging (MRI) is the method of choice for the diagnostics of focal and diffuse liver pathologies. In addition to its superior detection and differentiation capabilities, MRI can provide prognostic information and enable early assessment of the therapy response for malignant liver lesions using functional techniques, especially diffusion imaging. Computed tomography (CT) is the imaging method of choice for the detection of traumatic liver injury. Despite the increasing availability of functional techniques in CT, MRI remains the overall modality of choice in liver imaging.

The value of dynamic magnetic resonance imaging in interdisciplinary treatment of pelvic floor dysfunction.

PURPOSE: The purpose of this study was to determine the value of dynamic pelvic floor MRI relative to standard clinical examinations in treatment decisions made by an interdisciplinary team of specialists in a center for pelvic floor dysfunction. METHODS: 60 women were referred for dynamic pelvic floor MRI by an interdisciplinary team of specialists of a pelvic floor center. All patients were clinically examined by an urologist, gynecologist, a proctological, and colorectal surgeon. The specialists assessed individually and in consensus, whether (1) MRI provides important additional information not evident by physical examination and in consensus whether (2) MRI influenced the treatment strategy and/or (3) changed management or the surgical procedure. RESULTS: MRI was rated essential to the treatment decision in 22/50 cases, leading to a treatment change in 13 cases. In 12 cases, an enterocele was diagnosed by MRI but was not detected on physical exam. In 4 cases an enterocele and in 2 cases a rectocele were suspected clinically but not confirmed by MRI. In 4 cases, MRI proved critical in assessment of rectocele size. Vaginal intussusception detected on MRI was likewise missed by gynecologic exam in 1 case. CONCLUSION: MRI allows diagnosis of clinically occult enteroceles, by comprehensively evaluating the interaction between the pelvic floor and viscera. In nearly half of cases, MRI changed management or the surgical approach relative to the clinical evaluation of an interdisciplinary team. Thus, dynamic pelvic floor MRI represents an essential component of the evaluation for pelvic floor disorders.

[Value of new MR techniques in MR-PET]. / Stellenwert neuer MR-Techniken in der MR-PET.

The unparalleled soft tissue contrast of magnetic resonance imaging (MRI) and the functional information obtainable with 18-F fluorodeoxyglucose positron emission tomography (FDG-PET) render MR-PET well-suited for oncological and psychiatric imaging. The lack of ionizing radiation with MRI also makes MR-PET a promising modality for oncology patients requiring frequent follow-up and pediatric patients. Lessons learned with PET computed tomography (CT) over the last few years do not directly translate to MR-PET. For example, in PET-CT the Hounsfield units derived from CT are used for attenuation correction (AC). As 511 keV photons emitted in PET examinations are attenuated by the patient's body CT data are converted directly to linear attenuation coefficients (LAC); however, proton density measured by MRI is not directly related to the radiodensity or LACs of biological tissue. Thus, direct conversion to LAC data is not possible making AC more challenging in simultaneous MRI-PET scanning. In addition to these constraints simultaneous MRI-PET acquisitions also improve on some solutions to well-known challenges of hybrid imaging techniques, such as limitations in motion correction. This article reports on initial clinical experiences with simultaneously acquired MRI-PET data, focusing on the potential benefits and limitations of MRI with respect to motion correction as well as metal and attenuation correction artefacts.

DEPROMP Trial: the additive value of PSMA-PET/CT-guided biopsy for prostate cancer management in biopsy naïve men-study protocol for a randomized trial.

BACKGROUND: The primary objective is to determine the proportion of men with suspected prostate cancer (PCA) in whom the management plans are changed by additive gallium-68 prostate-specific membrane antigen positron emission tomography/computed tomography (PSMA-PET/CT) guided prostate biopsy (PET-TB) in combination with standard of care (SOC) using systematic (SB) and multiparametric magnetic resonance imaging-guided biopsy (MR-TB) compared with SOC alone. The major secondary objectives are to determine the additive value of the combined approach of SB + MR-TB + PET-TB (PET/MR-TB) for detecting clinically significant PCA (csPCA) compared to SOC; to determine sensitivity, specificity, positive and negative predictive value and diagnostic accuracy of imaging techniques, respective imaging classification systems, and each biopsy method; and to compare preoperatively defined tumor burden and biomarker expression and pathological tumor extent in prostate specimens. METHODS: The DEPROMP study is a prospective, open-label, interventional investigator-initiated trial. Risk stratification and management plans after PET/MR-TB are conducted randomized and blinded by different evaluation teams of experienced urologists based on histopathological analysis and imaging information: one including all results of the PET/MR-TB and one excluding the additional information gained by PSMA-PET/CT guided biopsy. The power calculation was centered on pilot data, and we will recruit up to 230 biopsy-naïve men who will undergo PET/MR-TB for suspected PCA. Conduct and reporting of MRI and PSMA-PET/CT will be performed in a blinded fashion. DISCUSSION: The DEPROMP Trial will be the first to evaluate the clinically relevant effects of the use of PSMA-PET/CT in patients with suspected PCA compared to current SOC. The study will provide prospective data to determine the diagnostic yields of additional PET-TB in men with suspected PCA and the impact on treatment plans in terms of intra- and intermodal changes. The results will allow a comparative analysis of risk stratification by each biopsy method, including a performance analysis of the corresponding rating systems. This will reveal potential intermethod and pre- and postoperative discordances of tumor stage and grading, providing the opportunity to critically assess the need for multiple biopsies. TRIAL REGISTRATION: German Clinical Study Register DRKS 00024134. Registered on 26 January 2021.

MR lymphangiography of lymphatic abnormalities in children and adults with Noonan syndrome.

Noonan syndrome is associated with complex lymphatic abnormalities. We report dynamic-contrast enhanced MR lymphangiography (DCMRL) findings in children and adults with Noonan syndrome to further elucidate this complex disease spectrum. A retrospective evaluation of patients with confirmed Noonan syndrome and clinical signs of lymphatic dysfunction undergoing DCMRL between 01/2019 and 04/2021 was performed. MRL included T2-weighted imaging (T2w) and DCMRL. Clinical history/presentation and genetic variants were recorded. T2w-imaging was evaluated for central lymphatic abnormalities and edema distribution. DCMRL was evaluated regarding the presence of cisterna chyli/thoracic duct, lymphatic leakages, pathological lymphatic reflux and abnormal lymphatic perfusion. The time from start of contrast-injection to initial enhancement of the thoracic duct venous junction was measured to calculate the speed of contrast propagation. Eleven patients with Noonan syndrome with lymphatic abnormalities (5 female, 6 male; 7 infants, 4 adults; mean age 10.8 ± 16.4 years) were identified (PTPN11 n = 5/11 [45.5%], RIT1 n = 5/11 [45.5%], KRAS n = 1/11 [9%]). Patients had a chylothorax (n = 10/11 [91%]) and/or pulmonary lymphangiectasia [dilated pulmonary lymph vessels] (n = 9/11 [82%]). Mediastinal/pulmonary edema was depicted in 9/11 (82%) patients. The thoracic duct (TD) was (partially) absent in 10/11 (91%) cases. DCMRL showed lymphatic reflux into intercostal (n = 11/11 [100%]), mediastinal (n = 9/11 [82%]), peribronchial (n = 8/11 [73%]), peripheral (n = 5/11 [45.5%]) and genital lymphatics (n = 4/11 [36%]). Abnormal pulmonary/pleural lymphatic perfusion was seen in 8/11 patients (73%). At infancy peripheral/genital edema was more prevalent in patients with RIT1 than PTPN11 (n = 3/5 vs. n = 0/5). Compared to patients with PTPN11 who had fast lymphatic enhancement in 4/5 patients, enhancement took markedly longer in 4/5 patients with RIT1-mutations. Thoracic duct dysplasia, intercostal reflux and pulmonary/pleural lymphatic perfusion are characteristic findings in patients with Noonan syndrome presenting with chylothorax and/or pulmonary lymphangiectasia. Central lymphatic flow abnormalities show possible phenotypical differences between PTPN11 and RIT1-mutations.

Ultrasound-guided needle positioning for nodal dynamic contrast-enhanced MR lymphangiography.

The aim of the study was to assess injection needle positioning for contrast-enhanced MR-lymphangiography (MRL) by ultrasound-guided injection of saline-solution. 80 patients (33 male, mean age 43.1 years) were referred for MRL. The injection needle position was assessed by injection of saline-solution. Consecutive lymph node distension was observed on sonography followed by MRL. Transpedal MRL was performed when no inguinal lymph nodes could be identified. The inguinal lymph node detection rate was recorded. MR-lymphangiograms were assessed regarding primary (i.e. enhancement of draining lymph vessels) and secondary technical success (i.e. lymph vessel enhancement after repositioning of the needle). MRL was considered as clinically successful if enhancement of the central lymphatic system and/or a lymphatic pathologies were observed. For a total of 92 MRLs 177 groins were evaluated sonographically. In 171/177 groins (96.6%) lymph nodes were identified. After needle placement lymph node distension was observed in 171/171 cases (100%) on saline injection. MR-contrast injection demonstrated enhancement of draining lymph vessels in 163/171 cases (95.3%). In 6/171 cases (3.5%) in-bore needle retraction lead to lymphatic enhancement. In one patient [2/171 nodes (1.1%)] no lymphatic enhancement was seen despite repeated needle repositioning. Overall contrast application was technically successful in 169/171 cases (98.8%). In the 6 groins in which no nodes were identifiable, transpedal MRL was successful. So overall 91/92 MRLs (98.9%) were clinically successful. No complications were recorded. Confirmation of the needle position for nodal MRL by sonographically controlled saline injection is a reliable technique with a high success rate of MRL.

"Number needed to read"--how to facilitate clinical trials in MR-angiography.

PURPOSE: To evaluate the effect of the number of readers on the statistical results in peripheral MRA. MATERIALS AND METHODS: 40 patients with peripheral arterial occlusive disease were included as a sample dataset in this study, randomly separated into two matched groups with n = 20 patients (group 1--gadobutrol, group 2--gadoterate meglumine) who underwent a continuous table movement MRA of the peripheral vessels at 3 T. Image quality (IQ) of 17 vessel segments was evaluated by 5 independent readers. The effect of the number of readers on significance and statistical power was statistically analyzed. RESULTS: Image quality in group 1 (gadobutrol) ranks significantly higher compared to group 2 (gadoterade meglumine) with a diagnostic IQ in 97% vs. 78% (p < 0.0001). For the diagnostic/non-diagnostic IQ assessment significance was reached with one reader 1/5 times (20%), with two readers in 4/10 (40%), with three readers in 6/10 (60%), with four readers in 4/5 (80%), with five readers in 1/1 (100%). Power considerations showed considerable gain when increasing the number of readers. CONCLUSION: Increasing the number of readers in a diagnostic MRA-study can be used to achieve a higher power or to decrease the number of subjects included with maintained statistical validity.

Evaluation of malignant effusions using MR-based T1 mapping.

Our aim was to investigate the diagnostic yield of rapid T1-mapping for the differentiation of malignant and non-malignant effusions in an ex-vivo set up. T1-mapping was performed with a fast modified Look-Locker inversion-recovery (MOLLI) acquisition and a combined turbo spin-echo and inversion-recovery sequence (TMIX) as reference. A total of 13 titrated albumin-solutions as well as 48 samples (29 ascites/pleural effusions from patients with malignancy; 19 from patients without malignancy) were examined. Samples were classified as malignant-positive histology, malignant-negative histology and non-malignant negative histology. In phantom analysis both mapping techniques correlated with albumin-content (MOLLI: r = - 0.97, TMIX: r = - 0.98). MOLLI T1 relaxation times were shorter in malignancy-positive histology fluids (2237 ± 137 ms) than in malignancy-negative histology fluids (2423 ± 357 ms) as well as than in non-malignant-negative histology fluids (2651 ± 139 ms); post hoc test for all intergroup comparisons: < 0.05. ROC analysis for differentiation between malignant and non-malignant effusions (malignant positive histology vs. all other) showed an (AUC) of 0.89 (95% CI 0.77-0.96). T1 mapping allows for non-invasive differentiation of malignant and non-malignant effusions in an ex-vivo set up.