Erratum: Author Correction: Advancing COVID-19 diagnosis with privacy-preserving collaboration in artificial intelligence.

[This corrects the article DOI: 10.1038/s42256-021-00421-z.].

Advancing COVID-19 Diagnosis with Privacy-Preserving Collaboration in Artificial Intelligence.

Artificial intelligence (AI) provides a promising substitution for streamlining COVID-19 diagnoses. However, concerns surrounding security and trustworthiness impede the collection of large-scale representative medical data, posing a considerable challenge for training a well-generalised model in clinical practices. To address this, we launch the Unified CT-COVID AI Diagnostic Initiative (UCADI), where the AI model can be distributedly trained and independently executed at each host institution under a federated learning framework (FL) without data sharing. Here we show that our FL model outperformed all the local models by a large yield (test sensitivity /specificity in China: 0.973/0.951, in the UK: 0.730/0.942), achieving comparable performance with a panel of professional radiologists. We further evaluated the model on the hold-out (collected from another two hospitals leaving out the FL) and heterogeneous (acquired with contrast materials) data, provided visual explanations for decisions made by the model, and analysed the trade-offs between the model performance and the communication costs in the federated training process. Our study is based on 9,573 chest computed tomography scans (CTs) from 3,336 patients collected from 23 hospitals located in China and the UK. Collectively, our work advanced the prospects of utilising federated learning for privacy-preserving AI in digital health.

Advancing COVID-19 diagnosis with privacy-preserving collaboration in artificial intelligence.

Artificial intelligence provides a promising solution for streamlining COVID-19 diagnoses; however, concerns surrounding security and trustworthiness impede the collection of large-scale representative medical data, posing a considerable challenge for training a well-generalized model in clinical practices. To address this, we launch the Unified CT-COVID AI Diagnostic Initiative (UCADI), where the artificial intelligence (AI) model can be distributedly trained and independently executed at each host institution under a federated learning framework without data sharing. Here we show that our federated learning framework model considerably outperformed all of the local models (with a test sensitivity/specificity of 0.973/0.951 in China and 0.730/0.942 in the United Kingdom), achieving comparable performance with a panel of professional radiologists. We further evaluated the model on the hold-out (collected from another two hospitals without the federated learning framework) and heterogeneous (acquired with contrast materials) data, provided visual explanations for decisions made by the model, and analysed the trade-offs between the model performance and the communication costs in the federated training process. Our study is based on 9,573 chest computed tomography scans from 3,336 patients collected from 23 hospitals located in China and the United Kingdom. Collectively, our work advanced the prospects of utilizing federated learning for privacy-preserving AI in digital health.

A collaborative online AI engine for CT-based COVID-19 diagnosis.

Artificial intelligence can potentially provide a substantial role in streamlining chest computed tomography (CT) diagnosis of COVID-19 patients. However, several critical hurdles have impeded the development of robust AI model, which include deficiency, isolation, and heterogeneity of CT data generated from diverse institutions. These bring about lack of generalization of AI model and therefore prevent it from applications in clinical practices. To overcome this, we proposed a federated learning-based Unified CT-COVID AI Diagnostic Initiative (UCADI, http://www.ai-ct-covid.team/), a decentralized architecture where the AI model is distributed to and executed at each host institution with the data sources or client ends for training and inferencing without sharing individual patient data. Specifically, we firstly developed an initial AI CT model based on data collected from three Tongji hospitals in Wuhan. After model evaluation, we found that the initial model can identify COVID from Tongji CT test data at near radiologist-level (97.5% sensitivity) but performed worse when it was tested on COVID cases from Wuhan Union Hospital (72% sensitivity), indicating a lack of model generalization. Next, we used the publicly available UCADI framework to build a federated model which integrated COVID CT cases from the Tongji hospitals and Wuhan Union hospital (WU) without transferring the WU data. The federated model not only performed similarly on Tongji test data but improved the detection sensitivity (98%) on WU test cases. The UCADI framework will allow participants worldwide to use and contribute to the model, to deliver a real-world, globally built and validated clinic CT-COVID AI tool. This effort directly supports the United Nations Sustainable Development Goals' number 3, Good Health and Well-Being, and allows sharing and transferring of knowledge to fight this devastating disease around the world.

Emergent spinal MRI in IVDU patients presenting with back pain: do we need an MRI in every case?

PURPOSE: Spinal MRI is the exam of choice for the workup of patients with suspected spinal infection. In this retrospective study, we assess the value of obtaining contrast-enhanced spinal MRI for patients presenting to the emergency department (ED) with acute back pain and a history of intravenous drug use (IVDU). METHODS: A retrospective IRB-approved, HIPAA compliant review of the imaging findings, reports and electronic charts of 167 consecutive IV drug-using patients (M/F = 96:71, mean age = 40 years) that presented to the ED with acute back pain over a 55-month period and underwent contrast-enhanced spinal MRI within 24 h. Fisher's exact test was used to identify statistically significant (p < 0.05) associations with MRI findings. RESULTS: Evidence of infectious spondylitis was demonstrated on the spinal MRIs of 39.5% (n = 66) of 167 patients, all of whom were admitted, and nearly half (48.5%; 32/66) underwent surgical or percutaneous intervention. Statistically significant differences in the decision to admit, blood cultures, and the type of treatment was demonstrated in patients with findings of spinal infection on MRI (p < 0.05). CONCLUSION: Use of emergent spinal MRI in the workup of IVDU patients with acute back pain is justified despite the resultant pressure on MRI scanner, technologist, and interpretation time.

Diffusion MRI of uterine and ovarian masses: identifying the benign lesions.

PURPOSE: The purpose of the study was to assess the diagnostic performance of qualitative and quantitative diffusion-weighted imaging (DWI) in differentiating benign from malignant ovarian and uterine masses. MATERIALS AND METHODS: Institutional review board approval was obtained for this HIPAA-compliant retrospective study, with waiver of informed consent. DWI MRIs of 222 women acquired over 1.5 years were evaluated. Reference standard was pathology or follow-up imaging. For qualitative assessment, two radiologists independently reviewed DWI and apparent diffusion coefficient (ADC) images for diffusion restriction. Differences were resolved by consensus. For quantitative assessment, a single reader measured ADC values. Readers were blinded to the reference standard. RESULTS: 222 lesions, 121 ovarian (99 benign and 22 malignant) and 101 uterine (54 benign and 47 malignant), were included. Final diagnosis was established with pathology in 129 (58%) or with imaging follow-up in 93 (42%). Mean (range) follow-up interval was 27 (13-48) months. Qualitative assessment yielded sensitivity (ratio, 95% CI), specificity, PPV and NPV of 100% (22/22, 85-100), 68% (68/99, 58-76), 41% (22/54, 27-54), and 100% (68/68, 94-100) for ovarian and 94% (44/47, 83-98), 91% (49/54, 80-96), 90% (44/49, 78-95) and 94% (49/52, 84-98) for uterine malignancies. ADC (mean ± SD) between benign ovarian [(1.11 ± 0.76) × 10-3 mm2/s] vs. malignant [(0.71 ± 0.26) × 10-3 mm2/s] lesions was significantly different (p < 0.001). ADC cutoff value of 1.55 × 10-3 mm2/s for ovarian lesions resulted in 99.9% confidence for the absence of malignancy. ADC (mean ± SD) of benign uterine [(0.64 ± 0.38) × 10-3 mm2/s] vs. malignant [(0.68 ± 0.19) × 10-3 mm2/s] lesions was not significantly different (P < 0.54). CONCLUSION: Quantitative and qualitative DWI assessment can be used to confidently characterize a subset of ovarian lesions as benign. With uterine lesions, although DWI is useful in differentiating benign from malignant lesions, the technique does not allow for definitive quantitative characterization.

Suboptimal CT pulmonary angiography in the emergency department: a retrospective analysis of outcomes in a large academic medical center.

Optimal CT pulmonary angiography (CTPA) is a prerequisite for accurate diagnosis and management of suspected venous thromboembolic disease (VTE) in the emergency department (ED). However, a certain proportion of CTPA studies are diagnostically limited or non-diagnostic due to various technical causes. In this study, we analyze the incidence and cause of suboptimal CTPA studies in the ED and assess the need for additional imaging. Reports of 1444 consecutive CTPAs performed in an ED on adult patients over a 25-month period beginning November 30, 2011, were reviewed. The observed suboptimal CTPA rate was 4.2 % (60/1444). The most common causes of limited or non-diagnostic CTPA in the ED were related to timing of contrast bolus or IV infiltration (26/60, 43.4 %), respiratory motion (16/60, 26.7 %), multifactorial causes (10/60, 16.7 %), and patient motion (8/60, 13.3 %). Of the 60 studies included, only 7 patients (11.7 %) underwent additional diagnostic imaging during the same hospital visit for VTE, while 3 patients (5.0 %) underwent additional imaging for suspected VTE over the next 2 months. A total of 2/60 (3.4 %) patients had documented acute PE on additional imaging performed either on the same hospital visit or within 2 months. Regardless of the factors contributing to suboptimal CTPA, only a very small proportion of patients receive additional imaging to evaluate for VTE, either on the same visit or during the next 2 months (16.7 %, 10/60 patients). A small number (3.4 %) of these patients have documented acute PE within 2 months when additional imaging tests were performed.

Radiology 24/7 In-House Attending Coverage: Do Benefits Outweigh Cost?

Many radiology practices, including academic centers, are moving to in-house 24/7 attending coverage. This could be costly and may not be easily accepted by radiology trainees and attending radiologists. In this article, we evaluated the effects of 24/7 in-house attending coverage on patient care, costs, and qualitative aspects such as trainee education. We retrospectively collected report turnaround times (TAT) and work relative value units (wRVU). We compared these parameters between the years before and after the implementation of 24/7 in-house attending coverage. The cost to provide additional attending coverage was estimated from departmental financial reports. A qualitative survey of radiology residents and faculty was performed to study perceived effects on trainee education. There were decreases in report TAT following 24/7 attending implementation: 69% reduction in computed tomography, 43% reduction in diagnostic radiography, 7% reduction in magnetic resonance imaging, and 43% reduction in ultrasound. There was an average daytime wRVU decrease of 9%, although this was compounded by a decrease in total RVUs of the 2013 calendar year. The financial investment by the institution was estimated at $850,000. Qualitative data demonstrated overall positive feedback from trainees and faculty in radiology, although loss of independence was reported as a negative effect. TAT and wRVU metrics changed with implementation of 24/7 attending coverage, although these metrics do not directly relate to patient outcomes. Additional clinical benefits may include fewer discrepancies between preliminary and final reports that may improve emergency and inpatient department workflows and liability exposure. Radiologists reported the impression that clinicians appreciated 24/7 in-house attending coverage, particularly surgical specialists. Loss of trainee independence on call was a perceived disadvantage of 24/7 attending coverage and raised a concern that residency education outcomes could be adversely affected.

Imaging of miscellaneous pancreatic pathology (trauma, transplant, infections, and deposition).

In this article's coverage of miscellaneous pancreatic topics, a brief review of pancreatic trauma; pancreatic transplantation; rare infections, such as tuberculosis; deposition disorders, including fatty replacement and hemochromatosis; cystic fibrosis; and others are discussed with pertinent case examples.

Evaluation of the effects of oral water and low-density barium sulphate suspension on bowel appearance on FDG-PET/CT.

The purpose of this study is to assess which of five bowel preparation regimes offers superior bowel distension and to assess if these regimes adversely affect FDG activity on PET/CT imaging. The study conformed to HIPAA regulations. Ninety patients were divided into five groups of 18 who received no oral contrast agent (group A); 900 ml of water orally (group B); or 900, 1,350, or 1,800 ml of LDB (groups C, D, E, respectively). PET/CT examinations were assessed quantitatively (bowel diameter, SUV) and qualitatively (visual assessment grading scale) for bowel distension and FDG activity by two blinded readers. ANOVA was utilized to determine if a statistically significant difference (SSD) existed between the groups in terms of distension and FDG uptake. Qualitatively superior bowel distension was observed in group C (LDB) compared to B (water) and greater distension was noted with increased volumes of LDB in C, D, and E. Quantitatively there was an SSD in mean distension between groups C and B (P < 0.001 except duodenum). Qualitatively and quantitatively there was no significant difference in bowel FDG uptake among the groups (P > 0.05). LDB as an oral contrast agent provides superior bowel distension over water and does not induce increased FDG bowel activity.

Imaging-based quantification of hepatic fat: methods and clinical applications.

Fatty liver disease comprises a spectrum of conditions (simple hepatic steatosis, steatohepatitis with inflammatory changes, and end-stage liver disease with fibrosis and cirrhosis). Hepatic steatosis is often associated with diabetes and obesity and may be secondary to alcohol and drug use, toxins, viral infections, and metabolic diseases. Detection and quantification of liver fat have many clinical applications, and early recognition is crucial to institute appropriate management and prevent progression. Histopathologic analysis is the reference standard to detect and quantify fat in the liver, but results are vulnerable to sampling error. Moreover, it can cause morbidity and complications and cannot be repeated often enough to monitor treatment response. Imaging can be repeated regularly and allows assessment of the entire liver, thus avoiding sampling error. Selection of appropriate imaging methods demands understanding of their advantages and limitations and the suitable clinical setting. Ultrasonography is effective for detecting moderate or severe fatty infiltration but is limited by lack of interobserver reliability and intraobserver reproducibility. Computed tomography allows quantitative and qualitative evaluation and is generally highly accurate and reliable; however, the results may be confounded by hepatic parenchymal changes due to cirrhosis or depositional diseases. Magnetic resonance (MR) imaging with appropriate sequences (eg, chemical shift techniques) has similarly high sensitivity, and MR spectroscopy provides unique advantages for some applications. However, both are expensive and too complex to be used to monitor steatosis.

Perfusion computed tomography for monitoring induction chemotherapy in patients with squamous cell carcinoma of the upper aerodigestive tract: correlation between changes in tumor perfusion and tumor volume.

OBJECTIVE: The aim of this study was to assess the potential of perfusion computed tomography (CTp) for monitoring induction chemotherapy in patients with squamous cell carcinoma (SCCA) of the upper aerodigestive tract. MATERIALS AND METHODS: Twenty-five patients with advanced SCCA underwent CTp and volumetric CT before and after induction chemotherapy. Perfusion CT parameters were calculated in the tumor, normal tissue, and muscles and correlated with tumor volume. RESULTS: The blood flow (BF), blood volume (BV), and permeability surface were significantly higher, and the mean transit time was significantly lower in the tumor than in the normal tissue. The tumor BF and BV significantly decreased, and the mean transit time significantly increased after the therapy; decrease in BF and BV correlated with tumor volume reduction after chemotherapy. The baseline tumor BV was significantly lower in nonresponders compared with that in responders. CONCLUSIONS: In patients with SCCA, CTp showed potential for monitoring induction chemotherapy, reduction in tumor BF and BV correlated with reduction of tumor volume after chemotherapy, and baseline tumor BV may predict response to chemotherapy.

Role of computed tomography perfusion in the evaluation of pancreatic necrosis and pancreatitis after endoscopic ultrasound-guided ablation of the pancreas in a porcine model.

OBJECTIVES: To evaluate the role of computed tomography (CT) perfusion in detection of pancreatic necrosis and pancreatitis after endoscopic ultrasound-guided ethanol ablation of porcine pancreas and to correlate the evaluation with histopathology. METHODS: Under endoscopic ultrasound guidance, 0.9% saline (control) and ethanol at 60%, 80%, and 100% concentrations were injected into the pancreatic tails of 4 pigs. On day 4, dynamic perfusion CT of the pancreas was performed. Perfusion analysis and evaluation of enhancement characteristics were done and correlated with histopathology. RESULTS: Ethanol injections at 80% and 100% concentrations resulted in focal necrosis surrounded by focal pancreatitis, whereas 60% ethanol injection caused severe focal pancreatitis with microscopic necrosis. The necrotic area revealed reduced blood flow, blood volume, permeability-surface area product, and increased mean transit time compared with pancreatitis and normal tissue (P < or = 0.001). In the control pig, no pancreatitis or necrosis was observed on perfusion images and histopathology. CONCLUSIONS: Pancreatic necrosis and pancreatitis after ethanol injection reduced the tissue perfusion on CT in comparison to normal tissue, with the changes being more substantial in necrosis than pancreatitis. These findings have possible implications in the accurate detection of pancreatic necrosis in patients with severe pancreatitis.

PET/CT for the characterization of adrenal masses in patients with cancer: qualitative versus quantitative accuracy in 150 consecutive patients.

OBJECTIVE: The objective of our study was to evaluate a large cohort of patients with PET/CT to determine whether qualitative (visual) assessment, quantitative standardized uptake value (SUV), or standardized uptake ratio (SUR) techniques should be used when attempting to characterize adrenal masses in patients with cancer. MATERIALS AND METHODS: The study group was composed of 150 consecutive patients (78 men, 72 women; mean age, 60 years; range, 24-88 years) with documented adrenal lesions. All patients were known to have an underlying primary malignancy and were referred for PET/CT to evaluate the underlying primary and metastatic tumor burden. Definitive lesion characterization was determined by evaluating all histologic adrenal specimens and all relevant prior and follow-up CT scans, including unenhanced, contrast-enhanced, and delayed contrast-enhanced washout studies. RESULTS: Of the 139 benign lesions, 109 were considered benign by CT densitometry measurements and 135 by qualitative PET data. Qualitative PET characterized 28 of 30 benign lesions that were considered indeterminate by unenhanced CT. All 26 malignant lesions were characterized by PET: All showed qualitative and quantitative signal intensity greater than the liver. By combining unenhanced and qualitative CT data with the retrospective PET data, the analysis yielded a sensitivity of 100% for the detection of malignancy, a specificity of 99%, a positive predictive value (PPV) of 93%, a negative predictive value (NPV) of 100%, and an accuracy of 99% (Table 1). Conversely, for the detection of benignity, the sensitivity, specificity, PPV, NPV, and accuracy were 99%, 100%, 100%, 93%, and 99%, respectively. CONCLUSION: PET/CT is a highly accurate method for differentiating benign from malignant adrenal masses particularly when using qualitative, rather than quantitative, PET data. The routine use of quantitative mean or maximal SUV or SUR data may be unnecessary. Occasional benign lesions do show mild to moderate increased FDG uptake compared with that of the liver and may mimic some malignant lesions. Without evidence that these lesions are benign by unenhanced CT densitometry or adrenal mass stability or growth from previous CT scans, we recommend that these lesions be characterized using contrast-enhanced washout tests and that if those tests are inconclusive, using percutaneous biopsy if early lesion characterization is mandatory.

Lymph node staging in esophageal adenocarcinoma with PET-CT based on a visual analysis and based on metabolic parameters.

BACKGROUND: In order to investigate the value of FDG positron emission tomography-computed tomography (PET-CT), FDG PET (reviewed side-by-side with CT), and metabolic parameters in the assessment of lymph node status and prognosis. METHODS: Fifty-five subjects with lymph node positive (N1) and 26 subjects with lymph node negative (N0) disease were included. In the slice with the maximum FDG uptake of the tumor, the axial area of the primary tumor, the maximum diameter of the tumor, and the mean and maximum standardized uptake values were measured. RESULTS: Fused PET-CT correctly characterized 289 of 325 lymph node groups (accuracy 89%) compared to 273 of 325 with PET (accuracy 84%). In lymph node staging (N0 vs. N1), PET-CT (accuracy 83%) was more accurate than PET (accuracy 78%). Among the metabolic parameters, the tumor diameter measured on PET-CT was the best predictor of lymph node stage (N0 vs. N1: accuracy 86%; threshold 25.5 mm) and overall survival. However, the highest accuracy of lymph node staging (N0 vs. N1) was achieved with the synergistic combination of visual analysis and primary tumor diameter measurements (accuracy 95%). CONCLUSIONS: PET-CT increases accuracy of lymph node staging in esophageal adenocarcinoma compared to PET. The primary tumor diameter further improves accuracy in lymph node staging and was shown to be an independent predictor of overall survival.

Issues in imaging malignant neoplasms of the female reproductive system.

Radiological evaluation of malignant neoplasms of the female reproductive system is invaluable in the initial diagnosis, staging, treatment planning, and follow-up management. Radiologists serving as consultants for the general primary care internist, gynecologists, and specialists in gynecologic oncology should be familiar with the strengths and limitations of various modalities used to evaluate gynecologic cancer patients. This article discusses the most common female reproductive tract neoplasms (ie, cervical, endometrial, and ovarian cancers) and the role of ultrasound, computed tomography, magnetic resonance imaging, and positron emission tomography in their diagnosis and management. Imaging features that impact on clinical diagnostic or treatment algorithms are highlighted and illustrated. Finally, recent technical advances that demonstrate promise are described.

Liver lesion detection and characterization in patients with colorectal cancer: a comparison of low radiation dose non-enhanced PET/CT, contrast-enhanced PET/CT, and liver MRI.

OBJECTIVES: To compare low-radiation dose non-enhanced fluorine 18 fluorodeoxyglucose (F-18 FDG) positron emission tomography (PET)/computed tomography (CT) (NE-PET/CT), contrast-enhanced fluorine 18 fluorodeoxyglucose PET/CT (CE-PET/CT), and gadolinium-enhanced liver magnetic resonance imaging (MRI) for the detection and characterization of liver lesions in patients with colorectal cancer (CRC). METHODS: In this retrospective review of imaging database of CRC patients with suspected liver metastases, 33 patients (22 men, 11 women; mean age, 63 years) evaluated with low-radiation dose NE-PET/CT, CE-PET/CT, and liver MRI were studied. The final diagnosis was established either by pathological examination or follow-up imaging over a period of at least 6 months for lesion stability or growth. The liver lesions were characterized on an ordinal scale of 0 to 6 (0 = absent, 1 = definitely benign, and 6 = definitely malignant). Receiver operating characteristic analysis was performed to compare performance of the 3 imaging methods. RESULTS: A total of 110 lesions were present on follow-up. The detection rate on low-radiation dose NE-PET/CT, CE-PET/CT, and MRI was 73.6%, 90.9%, and 95.4%, respectively. Magnetic resonance imaging (P < 0.001) and CE-PET/CT (P < 0.001) had a higher detection rate than low-radiation dose NE-PET/CT. There was no significant statistical difference in lesion detection between MRI and CE-PET/CT (P = 0.11). The sensitivity, specificity, and accuracy for characterization of detected liver lesions on low-radiation dose NE-PET/CT were 67%, 60%, and 66%, respectively; those on CE-PET/CT were 85%, 100%, and 86%, respectively; and those on MRI were 98%, 100%, and 98%, respectively. Comparative receiver operating characteristic analysis showed an area under curve of 0.74 for low-radiation dose NE-PET/CT, 0.86 for CE-PET/CT, and 0.97 for MRI. There were statistically significant differences in the accuracy of MRI, low-radiation dose NE-PET/CT, and CE-PET/CT for lesion characterization. CONCLUSIONS: When performing PET/CT, optimal detection and characterization of liver lesions require the use of a fused contrast-enhanced CT. Magnetic resonance imaging and CE-PET/CT have similar lesion detection rates. Magnetic resonance imaging is the best test for liver lesion characterization in patients with CRC.

Cholangiocarcinoma: current and novel imaging techniques.

The radiologic manifestations of cholangiocarcinomas are extremely diverse, since these tumors vary greatly in location, growth pattern, and histologic type. Familiarity with the imaging manifestations of cholangiocarcinomas is important for accurate detection and characterization of these tumors and assessment of resectability. Advances in imaging techniques have led to the availability of an array of modalities that, used independently or in combination, can aid in the accurate diagnosis and evaluation of cholangiocarcinomas in preparation for advanced surgical procedures and treatment planning. Response to novel targeted therapies can also be assessed with newer imaging tools. Hence, knowledge of current and emerging imaging applications is essential for correct diagnosis and appropriate management of these tumors.

Adenocarcinomas of the esophagus: response to chemoradiotherapy is associated with decrease of metabolic tumor volume as measured on PET-CT. Comparison to histopathologic and clinical response evaluation.

PURPOSE: We determined whether evaluation of treatment response is feasible by measuring metabolic tumor volume parameters on 18F-FDG (Fluorodeoxyglucose) PET-CT (Positron emission tomography-Computed tomography). We compared the response evaluation based on metabolic tumor volume parameters to a histopathologic and clinical response evaluation (clinical response criteria: RECIST criteria=Response evaluation criteria in solid tumors, and WHO criteria=World health organization). PATIENTS AND METHODS: A total of 51 study subjects with adenocarcinomas (Type I due to Siewert classification) of the esophagus underwent PET-CT scans before and after neoadjuvant chemoradiotherapy. Tumor volume, maximum and mean standardized uptake values (SUV) were assessed before and after chemoradiotherapy. Furthermore, the total lesion glycolysis (TLG) was calculated by multiplying the tumor volume by the mean SUV of the volume. Clinical response evaluation was performed with endoscopic ultrasound and CT using RECIST and WHO criteria. The reference standard for treatment response was the postsurgical histopathology. RESULTS: The decrease of tumor volume between the pre- and post-treatment PET-CT scans was a better predictor of histopathologic response and survival than the decrease of the SUV and of the clinical response evaluation based on RECIST and WHO criteria. The highest accuracy, however, was achieved when using the TLG for the identification of treatment responders. A decrease of the TLG by > 78% between pre- and post-therapy scans predicted histopathologic response with a sensitivity and specificity of 91% and 93%, respectively. CONCLUSIONS: Tumor volume and TLG can be used to assess treatment response and survival in patients with esophageal adenocarcinoma.

Optimal arterial phase imaging for detection of hypervascular hepatocellular carcinoma determined by continuous image capture on 16-MDCT.

OBJECTIVE: The purpose of this study is to estimate the optimal time delay before the initiation of arterial phase scanning for detection of hypervascular hepatocellular carcinoma (HCC) on 16-MDCT when a rapid bolus injection of contrast medium is administered. SUBJECTS AND METHODS: In this prospective study, 25 patients (19 men and six women; mean age, 63.5 years; age range, 50-81 years) with pathologically confirmed HCC were included. Dynamic 16-MDCT imaging was performed in cine mode using 70 mL of nonionic iodinated contrast medium (300 mg I/mL) at an injection rate of 7 mL/s. Four consecutive 5-mm-thick slices at the maximum diameter of the HCC were selected as the region of interest. Time-attenuation curves were generated by region of interest drawn on the aorta, tumor, and liver. Qualitative assessments of conspicuity for contrast medium wash-in, peak, and wash-out of aorta and tumor were performed. RESULTS: There were 108 arterial phase enhancing lesions (mean [+/-SD], 4.9 +/- 2.4 cm; range, 0.7-12.9 cm) in the 25 patients. The maximum Hounsfield value of aorta, tumor, and background liver parenchyma were 463.8 +/- 98 HU, 106.5 +/- 19 HU, and 98.3 +/- 14 HU, respectively. At the time of onset of peak tumor enhancement, the difference between tumor density and background liver density was 38.2 +/- 19 HU. The time-attenuation curve showed that the mean times of contrast enhancement start, peak, and end were 9.2 +/- 2.7 seconds, 19.4 +/- 2.1 seconds, and 38 +/- 13.5 seconds, respectively, for the aorta, and 15.5 +/- 2.6 seconds, 26.3 +/- 2.9 seconds, and 57.7 +/- 14.4 seconds, respectively, for 25 pathologically confirmed hepatocellular carcinomas. Qualitatively, the mean times of contrast enhancement wash-in, peak, and washout were 10.2 +/- 2.8 seconds, 19.9 +/- 3 seconds, and 39.9 +/- 9.2 seconds, respectively for the aorta, and 18 +/- 4.2 seconds, 27 +/- 3 seconds, and 55.7 +/- 21 seconds, respectively, for tumor. There were no differences between quantitative and qualitative measurements of wash-in and peak time for the aorta (p = 0.00017, p = 0.00016) and tumor (p = 0.00163, p = 0.00040). CONCLUSION: When using 70 mL of 300 mg I/mL of contrast medium with an injection rate of 7 mL/s in 16-MDCT scanning, the optimal time to initiate scanning for HCC is 26.3 +/- 2.9 seconds (range, 24.0-34.5 seconds) after contrast medium administration.