Natural Language Processing of Radiology Reports in Patients With Hepatocellular Carcinoma to Predict Radiology Resource Utilization.

OBJECTIVE: Radiology is a finite health care resource in high demand at most health centers. However, anticipating fluctuations in demand is a challenge because of the inherent uncertainty in disease prognosis. The aim of this study was to explore the potential of natural language processing (NLP) to predict downstream radiology resource utilization in patients undergoing surveillance for hepatocellular carcinoma (HCC). MATERIALS AND METHODS: All HCC surveillance CT examinations performed at our institution from January 1, 2010, to October 31, 2017 were selected from our departmental radiology information system. We used open source NLP and machine learning software to parse radiology report text into bag-of-words and term frequency-inverse document frequency (TF-IDF) representations. Three machine learning models-logistic regression, support vector machine (SVM), and random forest-were used to predict future utilization of radiology department resources. A test data set was used to calculate accuracy, sensitivity, and specificity in addition to the area under the curve (AUC). RESULTS: As a group, the bag-of-word models were slightly inferior to the TF-IDF feature extraction approach. The TF-IDF + SVM model outperformed all other models with an accuracy of 92%, a sensitivity of 83%, and a specificity of 96%, with an AUC of 0.971. CONCLUSIONS: NLP-based models can accurately predict downstream radiology resource utilization from narrative HCC surveillance reports and has potential for translation to health care management where it may improve decision making, reduce costs, and broaden access to care.

Which G-Tube to Use in Pullers: Assessment of Pull Pressures on Skin Models to Determine Optimal Catheter Choice in Patients with Recurrent Pulled Gastrostomy Tubes.

INTRODUCTION: Pulled or dislodged gastrostomy catheters represent a common complication associated with percutaneous gastrostomy and are a common cause of recurrent visits in patients with altered mental status. We intended to perform an experiment to compare the pull forces required to dislodge different commonly used gastrostomy catheters. MATERIALS AND METHODS: We used a digital force gauge device to measure the pull forces required to dislodge three types of 20 French gastrostomy catheters in double-layer skin models. These included the Flow 20 Pull Method (Cook Medical, Bloomington, IN, USA), Entuit Gastrostomy BR Balloon Retention feeding tube (Cook Medical, Bloomington, IN, USA), and Ponsky Non-Balloon Replacement Gastrostomy Tube (CR Bard Inc, Salt Lake City, Utah, USA). The catheters were inserted into the skin model using the same technique as would be utilized in a patient. RESULTS: The mean forces measured to dislodge the per-oral Flow 20 Pull Method, Entuit Thrive Balloon Retention, and button-type retention Ponsky replacement catheters were 35.6, 22.8, and 20.6 Newtons, respectively. The pull method per-oral gastrostomy catheter required significantly more pull force to dislodge than both the Ponsky button-type retention catheter and the Entuit balloon retention catheters. There was no significant difference in the pull force required to dislodge the Ponsky replacement catheter and the Entuit balloon retention catheter. CONCLUSIONS: Per-oral image-guided gastrostomy with pull-method button-type retention catheters may be the ideal choice in patients at high risk of tube dislodgment.

Minimally invasive fetal therapy for hydropic lung masses: three different approaches and review of the literature.

OBJECTIVE: To report three different antenatal therapeutic approaches for fetal lung masses associated with hydrops. METHODS: Three prospectively followed cases are described, and all 30 previously published minimally invasive cases of fetal therapy for hydropic lung masses are reviewed. RESULTS: Three hydropic fetuses with large intrathoracic lung masses presented at 17, 25 and 21 weeks of gestation, respectively. An aortic feeding vessel was identified in each case and thus a bronchopulmonary sequestration (BPS) was suspected. Under ultrasound guidance, the feeding vessel was successfully occluded with interstitial laser (Case 1), radiofrequency ablation (RFA) (Case 2) and thrombogenic coil embolization (Case 3). Complete (Cases 1 and 2) or partial (Case 3) resolution of the lung mass and hydrops was observed. A healthy infant was born at term after laser therapy (Case 1), and the involved lung lobe was resected on day 2 of postnatal life. In Case 2, hydrops resolved completely following RFA, but an iatrogenic congenital diaphragmatic hernia and abdominal wall defect became apparent 4 weeks later. The neonate died from sepsis following spontaneous preterm labor at 33 weeks. In Case 3, despite technical success in complete vascular occlusion with coils, a stillbirth ensued 2 days after embolization. CONCLUSIONS: The prognosis of large microcystic or echogenic fetal chest masses associated with hydrops is dismal. This has prompted attempts at treatment by open fetal surgery, with mixed results, high risk of premature labor and consequences for future pregnancies. We have demonstrated the possibility of improved outcome following ultrasound-guided laser ablation of the systemic arterial supply. Despite technical success, RFA and coil embolization led to procedure-related complications and need further evaluation.

Multidisciplinary Canadian consensus recommendations for the management and treatment of hepatocellular carcinoma.

Globally, hepatocellular carcinoma (hcc) is the third most common cause of death from cancer, after lung and stomach cancer. The incidence of hcc in Canada is increasing and is expected to continue to increase over the next decade. Given the high mortality rate associated with hcc, steps are required to mitigate the impact of the disease. To address this challenging situation, a panel of 17 hcc experts, representing gastroenterologists, hepatologists, hepatobiliary surgeons, medical oncologists, pathologists, and radiologists from across Canada, convened to provide a framework that, using an evidence-based approach, will assist clinicians in optimizing the management and treatment of hcc. The recommendations, summarized here, were developed based on a rigorous methodology in a pre-specified process that was overseen by the steering committee. Specific topics were identified by the steering committee and delegated to a group of content experts within the expert panel, who then systematically reviewed the literature on that topic and drafted the related content and recommendations. The set of recommendations for each topic were reviewed and assigned a level of evidence and grade according to the levels of evidence set out by the Centre for Evidence-based Medicine, Oxford, United Kingdom. Agreement on the level of evidence for each recommendation was achieved by consensus. Consensus was defined as agreement by a two-thirds majority of the 17 members of the expert panel. Recommendations were subject to iterative review and modification by the expert panel until consensus could be achieved.

Clinical utility of ultrasound versus magnetic resonance imaging for deciding to proceed with uterine artery embolization for presumed symptomatic fibroids.

AIM: To compare the diagnostic utility of pelvic ultrasound (US) and magnetic resonance imaging (MRI) on the clinical decision to proceed with uterine artery embolization (UAE). MATERIALS AND METHODS: Over 2 years, 180 consecutive women (mean age 43) sought consultation for UAE, 116 underwent pelvic US and MRI before possible UAE. US was performed prior to MRI. Imaging was analysed for leiomyoma quantity, size and location, uterine volume, and the presence of potential contraindications to UAE. Discrepancies between imaging methods and cases where discrepancies could have altered management, were recorded. RESULTS: For the 116 patients who completed imaging, the average uterine volume was 701 cm(3) using MRI versus 658 cm(3) using US (p=0.48). The average dominant leiomyoma volume was 292 cm(3) using MRI versus 253 cm(3) using US (p=0.16). In 14 (12.1%) patients US did not correctly quantify or localize leiomyomas compared with MRI (p=0.0005). Thirteen patients did not undergo UAE (patient preference n=9, pre-procedural imaging findings n=4). In the four cases where UAE was not performed due to imaging findings, relevant findings were all diagnosed by MRI compared with two by US (p=0.5). The two cases not detected by ultrasound were adenomyosis and a pedunculate subserosal leiomyoma. Of the 103 patients who underwent UAE, 14 were treated (without complication) despite the presence of a relative contraindication; all 14 relative contraindications were identified by MRI compared with 13 by US (p=1.0). CONCLUSION: MRI is more accurate than US for characterizing uterine leiomyomas. In a small but statistically insignificant number of cases, MRI identified findings that were missed by US, which changed management. For patients that are unsuitable to be assessed with MRI, ultrasound alone is sufficient for pre-UAE assessment.

Percutaneous ethanol injection of unresectable medium-to-large-sized hepatomas using a multipronged needle: efficacy and safety.

Fine needles with an end hole or multiple side holes have traditionally been used for percutaneous ethanol injection (PEI) of hepatomas. This study retrospectively evaluates the safety and efficacy of PEI of unresectable medium-to-large (3.5-9 cm) hepatomas using a multipronged needle and with conscious sedation. Twelve patients, eight men and four women (age 51-77 years; mean: 69) received PEI for hepatomas, mostly subcapsular or exophytic in location with average tumor size of 5.6 cm (range: 3.5-9.0 cm). Patients were consciously sedated and an 18G retractable multipronged needle (Quadrafuse needle; Rex Medical, Philadelphia, PA) was used for injection under real-time ultrasound guidance. By varying the length of the prongs and rotating the needle, the alcohol was widely distributed within the tumor. The progress of ablation was monitored by contrast-enhanced ultrasound, computed tomography (CT) or magnetic resonance imaging (MRI) after each weekly injection and within a month after the final (third) injection and 3 months thereafter. An average total of 63 mL (range: 20-154 ml) of alcohol was injected per patient in an average of 2.3 sessions. Contrast-enhanced CT, ultrasound, or MRI was used to determine the degree of necrosis. Complete necrosis was noted in eight patients (67%), near-complete necrosis (90-99%) in two (16.7%), and partial success (50-89%) in two (16.7%). Follow-up in the first 9 months showed local recurrence in two patients and new lesions in another. There was no mortality. One patient developed renal failure, liver failure, and localized perforation of the stomach. He responded to medical treatment and surgery was not required for the perforation. One patient had severe postprocedural abdominal pain and fever, and another had transient hyperbilirubinemia; both recovered with conservative treatment. PEI with a multipronged needle is a new, safe, and efficacious method in treating medium-to-large-sized hepatocellular carcinoma under conscious sedation. Its survival benefits require further investigations.

Transarterial chemoembolisation for advanced hepatocellular carcinoma: results from a North American cancer centre.

AIMS: In Asian countries, transarterial chemoembolisation (TACE) has long been used for palliation of unresectable hepatocellular carcinoma (HCC) without strong evidence of improved survival or quality of life. In 2002, a survival benefi of TACE was shown in two randomised controlled trials in Europe and Hong Kong. The effectiveness of interventions fo HCC is influenced by geographical factors related to diverse patient characteristics and protocols. Therefore, the validation of TACE as palliative modality for unresectable HCC requires confirmation in diverse patient populations. The aim of the present study was to assess the effectiveness of TACE for HCC in a North American population. MATERIALS AND METHODS: This was a single centre prospective cohort study. Child-Pugh A cirrhosis or better patients wit unresectable HCC and without radiological evidence of metastatic disease or segmental portal vein thrombosis wer assessed between November 2001 and May 2004. Of 54 patients who satisfied the inclusion criteria, 47 underwent 80 TACE sessions. Chemoembolisation was carried out using selective hepatic artery injection of 75 mg/m(2) doxorubicin and lipiodol followed by an injection of embolic particles when necessary. Repeat treatments were carried out at 2-3 month intervals for recurrent disease. The primary outcome was overall survival; secondary outcomes were morbidity and tumour response. RESULTS: The survival probabilities at 1, 2 and 3 years were 76.6, 55.5 and 50%, respectively. At 6 months after the first intervention, 31% of patients had a partial response and 60% had stable disease by RECIST criteria. Minor adverse events occurred after 39% of TACEs and major adverse events after 20% of sessions, including two treatment-related deaths (4% of patients). One patient had complete cancer remission after undergoing three TACE treatments. Further progression of tumour growth was prevented in 91% of tumours at the 6 month point after the first TACE. At 3 months, serum levels of the tumour marker alpha-feto protein were significantly reduced in patients with elevated levels before TACE. CONCLUSIONS: The survival probabilities at 1 and 2 years after TACE were comparable with results in randomised studies from Europe and Asia. Most patients tolerated TACE well, but clinicians need to be aware that moderately severe sideeffects require close monitoring and prompt intervention.

Assessing damage in individual joints in rheumatoid arthritis: a new method based on the Larsen system.

OBJECTIVES: To evaluate observer agreement using the Larsen system (LS) and a Modified Larsen system (ML) when assessing individual joints of the hands and wrists in rheumatoid arthritis, and to compare the two systems. To determine the minimally important difference (MID) for the ML. METHODS: Thirty radiographs of hands and wrists from 10 patients who presented with RA were graded by two blinded observers, using the LS and then the ML. Patients were followed for a mean of 7.2 years (range: 4-10 years). Inter- and intra-observer agreement were calculated using the kappa statistic with linear incremental weights. Inter-observer agreement was also computed for the summed score, using an intraclass correlation coefficient. Inter-observer error was estimated by calculating the mean and standard deviation of the grading differences between the two observers. Prevalence of damage was calculated as a ratio of damage: no damage and expressed as a percentage. Pairs of radiographs were comparatively graded using a seven-point Likert scale. RESULTS: The kappa statistic for inter-observer agreement was 0.38 (marginal reproducibility) for the LS and 0.52 (good reproducibility) for the ML (P = 0.004). Using a difference of one grade as perfect agreement, it was 0.56 (good reproducibility) for the LS and 0.87 (excellent reproducibility) for the ML (P = 0.001). Intra-observer agreement was high in both systems. The distribution of ML-grade differences varied according to the level of the Likert scale: for "a little bit worse", representing the smallest amount of detectable damage progression, the distribution differences peaked around two grades. This value represented a MID 87% of the time. CONCLUSIONS: The LS lacks precision for individual joints. The ML, it is proposed, has more detailed definitions of grades, and is more reliable. When pairs of radiographs were compared, a two-grade difference on the ML was the MID.

Günther Tulip Retrievable Vena Cava Filter: results from the Registry of the Canadian Interventional Radiology Association.

PURPOSE: To report data collected by the Canadian Registry of the Günther Tulip Retrievable Filter (GTF). MATERIALS AND METHODS: Between February 1998 and December 2000, 90 patients at eight hospitals underwent implantation of 91 GTFs. There were 45 male patients and 45 female patients, age 17-88 years, with a mean age of 49 years. Indications for filter placement were pulmonary embolism (PE) or deep vein thrombosis (DVT) with a contraindication to anticoagulation in 83 patients, prophylaxis after massive PE in one, prophylaxis for proximal free-floating thrombus in one, and prophylaxis with no DVT or PE in six patients (major trauma, n = 4; high preoperative risk, n = 2). GTF retrieval was attempted in selected patients from a right internal jugular vein approach. RESULTS: One GTF was inadvertently placed in the right iliac vein and could not be retrieved. There were no other major placement complications. GTF retrieval was attempted in 52 patients (53 GTFs); 52 GTFs were successfully retrieved from 51 patients. Implantation times were 2-25 days (mean, 9 d). Of these 51 patients, 37 underwent follow-up for 5-420 days (mean, 103 d) after filter retrieval. Four patients (8% of retrieved GTFs) required reinsertion of a permanent filter 17-167 days (mean, 78 d) after GTF retrieval as a result of bleeding from anticoagulation (n = 2) or because the patient required further surgery (n = 2). One other patient had recurrent DVT 230 days after retrieval; no PE or other complication was documented in the retrieval group. GTFs were not retrieved from 39 patients for various reasons. Of these 39 patients, 25 underwent follow-up 7-420 days (mean, 85 d) after filter placement. Two patients developed filter occlusion (5%); no other complications were documented. CONCLUSION: The GTF has a broad range of utility: it can be used as a permanent filter or retrieved after implantation periods of 15 days and possibly longer. However, indications for retrieval require further study, as does the maximum implantation time.

Development of a radiofrequency based thermal therapy technique in an in vivo porcine model for the treatment of small renal masses.

PURPOSE: Incidentally detected small renal tumors appear to grow slowly and be localized to the kidney. Minimally invasive therapies are being investigated as alternatives to standard surgical techniques. Radiofrequency ablation has been reported for the treatment of small renal cell carcinomas. We developed a radiofrequency technique and established its efficacy and safety in a large animal model. METHODS AND METHODS: A total of 22 lesions were created in normal kidneys of 7 pigs. Radiofrequency energy was administered during open exposure of the kidneys or percutaneously under ultrasound guidance. Lesion development was monitored with gray-scale and power Doppler ultrasound. To avoid heating surrounding tissues new hydro-dissection and gas-dissection techniques were developed. Lesion sizes and characteristics were assessed by ultrasound and pathological examination. RESULTS: No complications were observed due to probe insertion and removal. Perirenal structures were thermally damaged before the development and application of the dissection techniques. Lesion size was accurately predicted by gray-scale ultrasound on day 7. Loss of perfusion in the ablated volume was confirmed by power Doppler ultrasound. Lesions were wedge-shaped, presumably due to the effects of heating on segmental blood flow distribution. Pathological examination revealed changes consistent with thermal injury and ischemic type infarction. CONCLUSIONS: Radiofrequency thermal therapy is an effective and efficient method for ablating normal renal tissue in the pig. It may be applied percutaneously under ultrasound guidance with minimal complications provided that vital adjacent structures are protected from thermal damage. Further studies are required in humans before adopting this technique as definitive treatment for small renal cell carcinoma.

Agreement between alternative classifications of acute respiratory distress syndrome.

To examine the agreement between two classifications of acute respiratory distress syndrome (ARDS) that are used interchangeably in clinical practice and clinical research, we classified 118 patients taking part in a randomized trial with respect to the presence of ARDS using the North American-European Consensus Committee (NAECC) and the Lung Injury Severity Score (LISS) criteria. The incidence of ARDS using NAECC criteria was 55.1% (95% confidence interval, 46.1% to 64.1%), and using the LISS criteria 61.9% (95% confidence interval, 53.1% to 70.6%). The p value on the difference between these proportions was 0.07. Raw agreement, chance-corrected agreement (kappa), and chance-independent agreement (phi) on the study occurrence of ARDS using the two classifications were, respectively, 0.73 (95% CI, 0.65 to 0.81), 0.46 (95% CI, 0.32 to 0.61), and 0.63 (95% CI, 0.41 to 0.79). No single component of either index contributed to disagreement to an appreciably greater extent than other components. Baseline characteristics and outcomes were similar among patients who developed ARDS according to either classification. We conclude that NAECC and LISS classifications resulted in similar estimates of the incidence of ARDS in this clinical trial, though patients were frequently classified as having ARDS with only one model. These discordant classifications had no prognostic importance.

Interobserver variation in interpreting chest radiographs for the diagnosis of acute respiratory distress syndrome.

To measure the reliability of chest radiographic diagnosis of acute respiratory distress syndrome (ARDS) we conducted an observer agreement study in which two of eight intensivists and a radiologist, blinded to one another's interpretation, reviewed 778 radiographs from 99 critically ill patients. One intensivist and a radiologist participated in pilot training. Raters made a global rating of the presence of ARDS on the basis of diffuse bilateral infiltrates. We assessed interobserver agreement in a pairwise fashion. For rater pairings in which one rater had not participated in the consensus process we found moderate levels of raw (0.68 to 0.80), chance-corrected (kappa 0.38 to 0.55), and chance-independent (Phi 0. 53 to 0.75) agreement. The pair of raters who participated in consensus training achieved excellent to almost perfect raw (0.88 to 0.94), chance-corrected (kappa 0.72 to 0.88), and chance-independent (Phi 0.74 to 0.89) agreement. We conclude that intensivists without formal consensus training can achieve moderate levels of agreement. Consensus training is necessary to achieve the substantial or almost perfect levels of agreement optimal for the conduct of clinical trials.

The Simmen classification of wrist destruction in rheumatoid arthritis. Experience in patients with early disease.

In this radiological study, bilateral radiographs of the wrist in 48 patients with early rheumatoid arthritis (RA) were graded by the Simmen classification, at different times, to determine its potential use in early disease. Interobserver agreement was good. The Simmen classification of wrist destruction in RA could be applied to the radiographs of approximately 50% of patients with early disease. It has some consistency over time.

Long-term follow-up of survivors of acute lung injury: lack of effect of a ventilation strategy to prevent barotrauma.

OBJECTIVES: To determine the effect of a ventilation strategy to prevent barotrauma on long-term outcome in survivors of acute lung injury. DESIGN: Prospective blinded cohort analysis. SETTING: Three university-affiliated medical-surgical intensive care units. PATIENTS: A total of 28 survivors of acute lung injury, 1-2 yrs after diagnosis, from a multicenter prospective randomized controlled trial comparing pressure (peak inflation pressure < or =30 cm H2O) and volume (tidal volume < or =8 mL/kg) limited ventilation to a conventional (peak inflation pressure < or =50 cm H2O, tidal volume 10-15 mL/kg) ventilation strategy. MEASUREMENTS AND MAIN RESULTS: Physicians blinded as to treatment group evaluated 20 of 28 survivors (treatment group, 7; control group, 13). Exercise tolerance in the 6-minute walk test was comparable to patients with chronic respiratory disease and equivalent between groups (treatment group, 373+/-171 m vs. control group, 375+/-129 m; p = .84). Pulmonary function testing showed reduced diffusing capacity (treatment group, 64+/-29% predicted vs. control group, 74+/-14% predicted; p = .68) and normal volumes, flows, and blood gases. Two domains of disease-specific Health Related Quality of Life assessed by the Chronic Respiratory Questionnaire were worse for patients in the treatment group compared with the control group (Emotional Function 3.8+/-1.4 vs. 5.1+/-0.08; p = .05, Mastery 4.7+/-1.7 vs. 6.2+/-0.8; p = .03). There were no between-group differences in the scores of the Spitzer Quality of Life Index (a generic Health Related Quality of Life instrument), although they were reduced (7.5+/-1.9) and comparable to patients with chronic disease. CONCLUSIONS: We found that 1-2 yrs after the onset of their illness, survivors of acute lung injury have reductions in quality of life and exercise tolerance which are similar to patients with chronic diseases. We were unable to show that a limited ventilation strategy improves either long-term pulmonary function or quality of life in survivors of acute lung injury.