Real-World Performance of Large Vessel Occlusion Artificial Intelligence-Based Computer-Aided Triage and Notification Algorithms-What the Stroke Team Needs to Know

PURPOSE

To evaluate the real-world performance of two FDA-approved artificial intelligence (AI)-based computer-aided triage and notification (CADt) detection devices and compare them with the manufacturer-reported performance testing in the instructions for use.

MATERIALS AND METHODS

Clinical performance of two FDA-cleared CADt large-vessel occlusion (LVO) devices was retrospectively evaluated at two separate stroke centers. Consecutive "code stroke" CT angiography examinations were included and assessed for patient demographics, scanner manufacturer, presence or absence of CADt result, CADt result, and LVO in the internal carotid artery (ICA), horizontal middle cerebral artery (MCA) segment (M1), Sylvian MCA segments after the bifurcation (M2), precommunicating part of cerebral artery, postcommunicating part of the cerebral artery, vertebral artery, basilar artery vessel segments. The original radiology report served as the reference standard, and a study radiologist extracted the above data elements from the imaging examination and radiology report.

RESULTS

At hospital A, the CADt algorithm manufacturer reports assessment of intracranial ICA and MCA with sensitivity of 97% and specificity of 95.6%. Real-world performance of 704 cases included 79 in which no CADt result was available. Sensitivity and specificity in ICA and M1 segments were 85.3% and 91.9%. Sensitivity decreased to 68.5% when M2 segments were included and to 59.9% when all proximal vessel segments were included. At hospital B the CADt algorithm manufacturer reports sensitivity of 87.8% and specificity of 89.6%, without specifying the vessel segments. Real-world performance of 642 cases included 20 cases in which no CADt result was available. Sensitivity and specificity in ICA and M1 segments were 90.7% and 97.9%. Sensitivity decreased to 76.4% when M2 segments were included and to 59.4% when all proximal vessel segments are included.

DISCUSSION

Real-world testing of two CADt LVO detection algorithms identified gaps in the detection and communication of potentially treatable LVOs when considering vessels beyond the intracranial ICA and M1 segments and in cases with absent and uninterpretable data.

Korrelieren szintigraphisch ermittelte Knochenstoffwechselaktivitäten mit der Kalzium- und Phosphorbestimmung durch die Röntgen-Mikroanalyse?

An animal experiment on bone healing of dental implants offered the opportunity of comparing scintigraphic images with histological findings, specially with electron microscope results and analysis of the calcium and phosphorus values in different stages. The clear correspondence between the scintigraphic results and the calcification level of the new bone parts must be seen with reservation because of possible methodical mistakes of the radio-micro-analysis from surfaces and of intraindividual differences.

Collaborative Interventions Reduce Time-to-Thrombolysis for Acute Ischemic Stroke in a Public Safety Net Hospital

BACKGROUND AND PURPOSE

Shorter time-to-thrombolysis in acute ischemic stroke (AIS) is associated with improved functional outcome and reduced morbidity. We evaluate the effect of several interventions to reduce time-to-thrombolysis at an urban, public safety net hospital.

METHODS

All patients treated with tissue plasminogen activator for AIS at our institution between 2008 and 2015 were included in a retrospective analysis of door-to-needle (DTN) time and associated factors. Between 2011 and 2014, we implemented 11 distinct interventions to reduce DTN time. Here, we assess the relative impact of each intervention on DTN time.

RESULTS

The median DTN time pre- and postintervention decreased from 87 (interquartile range: 68-109) minutes to 49 (interquartile range: 39-63) minutes. The reduction was comprised primarily of a decrease in median time from computed tomography scan order to interpretation. The goal DTN time of 60 minutes or less was achieved in 9% (95% confidence interval: 5%-22%) of cases preintervention, compared with 70% (58%-81%) postintervention. Interventions with the greatest impact on DTN time included the implementation of a stroke group paging system, dedicated emergency department stroke pharmacists, and the development of a stroke code supply box.

CONCLUSIONS

Multidisciplinary, collaborative interventions are associated with a significant and substantial reduction in time-to-thrombolysis. Such targeted interventions are efficient and achievable in resource-limited settings, where they are most needed.

Abstract TP241: High-yield Interventions Reduce Time to tPA in Ischemic Stroke

Introduction: Tissue plasminogen activator (tPA) reduces disability from ischemic stroke, but this benefit is time dependent. Guidelines recommend a door-to-tPA time of < 60 mins. We retrospectively evaluated the effect of eight interventions on door-to-tPA time at a large safety net public hospital.

Hypothesis: High-yield interventions would significantly reduce door-to-tPA times, especially a stroke code activation system.

Methods: From Jan 2008 to Dec 2014, all patients (n=537) presenting to our emergency department (ED) with possible stroke and for whom tPA was considered were entered into a database. Critical ischemic stroke times were recorded and compared annually (physician at bedside, labs resulted, CT scan read, tPA given). Interventions to reduce time to tPA were introduced intermittently (Figure). Mean time to tPA was compared for the six-month period before and after each intervention was implemented. Student’s t-test was used with significance at p<0.05.

Results: The mean time to tPA improved from 2008 (98.9 min) to 2014 (60.0 min; p<0.001). This was largely driven by reductions in mean times from ED physician evaluation to head CT read (-35.3 min; p<0.0001) and from ED physician to neurologist evaluation (-23.3 min; p<0.0001). The interventions associated with the greatest reductions in mean time to tPA were an ED stroke pharmacist/giving tPA in the CT scanner (-17.5 min) and conducting monthly multidisciplinary stroke peer reviews (-15.9 min).

Conclusions: A safety net public hospital significantly reduced its time to tPA for stroke through a combination of interventions, including multidisciplinary stroke peer reviews and infusing tPA in the CT scanner.

Abstract TP311: Multidisciplinary Stroke Simulation: Novel Team Based Approach to Decrease Door-to-Needle Times

Introduction: In acute stroke, time equals brain. Minimizing time to treatment maximizes eligibility and effectiveness of fibrinolytics. Timely treatment of acute stroke requires precise coordination of a multidisciplinary team ranging from first responders to neurointerventionalists. Simulation-based learning allows participants to hone their skills and make mistakes in a controlled environment. To our knowledge, the impact of multidisciplinary stroke simulation has not been reported in literature. Here we describe our initial experience implementing such a simulation.

Hypothesis: Participation in stroke simulation will improve knowledge of acute stroke care guidelines and decrease door-to-needle time.

Methods: Neurology, emergency medicine and radiology trainees, EMTs, nurses, medical students, technologists, and pharmacists took part in the evaluation and treatment of simulated patients with stroke symptoms in actual clinical settings from ambulance to ED to CT scanner to IR suite. Neurology and neuroradiology faculty debriefed participants following simulations. Questions on stroke care (derived from the 2013 AHA/ASA guideline and 2015 update) were sent to likely participants before and after the simulation; those who completed pre/post quizzes and the simulation were included in analysis.

Results: Survey response rate was 86%. All participants had improved scores on the post-simulation quiz, scoring an average of 19% higher, 95% CI [8%, 29%]. For example, correct responses that IV tPA is not contraindicated prior to endovascular therapy improved from 64 to 100% after the simulation; responses correctly identifying the appearance of ischemic penumbra on CT perfusion imaging increased from 27 to 73%. Nearly all (92%) respondents would recommend the simulation to their peers.

Conclusions: Simulation of acute stroke scenarios improves participants’ knowledge of acute stroke management guidelines and may improve door-to-needle time. We present a novel framework for multidisciplinary simulation, which could be implemented at other institutions. Further evaluation of simulation effect on door-to-needle time is ongoing.

Medical Student Usage of the American College of Radiology Appropriateness Criteria

RATIONALE AND OBJECTIVES

Educating medical students on appropriate imaging utilization has been increasingly recognized as important for patient care. The American College of Radiology Appropriateness Criteria (ACR-AC) is designed to support evidence-based imaging examination selection. We sought to assess whether medical students order imaging studies independently, what resources they use for guidance, and whether they use the ACR-AC in clinical practice. A secondary aim was to determine whether increasing familiarity with the ACR-AC could impact student usage.

MATERIALS AND METHODS

We surveyed third year medical students at a single institution on their imaging practices, familiarity with the ACR-AC, and preferences among available resources to guide proper examination selection. The survey was performed in person before a lecture. We also designed a brief intervention to improve familiarity with the ACR-AC and then reassessed students to determine any effect on utilization.

RESULTS

The response rate for the initial survey was 103 of 109 (94%) and the response rate for the second survey was 99 of 109 (91%).Our initial survey found students initiated imaging orders independently (74 of 100, 74.8%) and consulted resources to assist in examination selection (50 of 74, 67.6%). Students expressed a preference for non-ACR-AC resources, notably Up to Date via its online mobile application.Few students (8 of 71, 11.3%) were familiar with the ACR-AC. After an intervention to increase familiarity with the ACR-AC, student awareness of the ACR-AC increased to 61 of 74 (82.4%). However, usage among those familiar with the resource remained low, 13 of 61(21.3%) versus 3 of 8 (37.5%).

CONCLUSIONS

Use of the ACR-AC was low among third year medical students. After increasing students' familiarity with the ACR-AC, their usage in a clinical setting did not increase. The largest barrier to use may be the lack of a quick, easy to use online mobile application-based interface.

Rosai-Dorfman Disease Isolated to the Thoracic Epidural Spine

Rosai-Dorfman disease is a rare benign histiocytic disease that infrequently presents in the spine. We report a case of Rosai-Dorfman disease isolated to the epidural thoracic spine in a 26-year-old male. To our knowledge, this is the 15th reported case of isolated spinal disease and only the fourth case of isolated thoracic epidural disease. Given its rarity as well as non-specific symptoms and imaging findings, Rosai-Dorfman disease is often not considered and misdiagnosed on imaging studies. To help improve awareness of Rosai-Dorfman spinal disease, we review the literature and discuss the epidemiology, clinical presentation, imaging features, and treatment considerations for this condition.

Katanin localization requires triplet microtubules in Chlamydomonas reinhardtii

Centrioles and basal bodies are essential for a variety of cellular processes that include the recruitment of proteins to these structures for both centrosomal and ciliary function. This recruitment is compromised when centriole/basal body assembly is defective. Mutations that cause basal body assembly defects confer supersensitivity to Taxol. These include bld2, bld10, bld12, uni3, vfl1, vfl2, and vfl3. Flagellar motility mutants do not confer sensitivity with the exception of mutations in the p60 (pf19) and p80 (pf15) subunits of the microtubule severing protein katanin. We have identified additional pf15 and bld2 (ε-tubulin) alleles in screens for Taxol sensitivity. Null pf15 and bld2 alleles are viable and are not essential genes in Chlamydomonas. Analysis of double mutant strains with the pf15-3 and bld2-6 null alleles suggests that basal bodies in Chlamydomonas may recruit additional proteins beyond katanin that affect spindle microtubule stability. The bld2-5 allele is a hypomorphic allele and its phenotype is modulated by nutritional cues. Basal bodies in bld2-5 cells are missing proximal ends. The basal body mutants show aberrant localization of an epitope-tagged p80 subunit of katanin. Unlike IFT proteins, katanin p80 does not localize to the transition fibers of the basal bodies based on an analysis of the uni1 mutant as well as the lack of colocalization of katanin p80 with IFT74. We suggest that the triplet microtubules are likely to play a key role in katanin p80 recruitment to the basal body of Chlamydomonas rather than the transition fibers that are needed for IFT localization.

[Inhibition of 18F-FDG uptake in glioblastoma cells by FDG and glucose]

It was the aim of the study to compare the inhibition of 18F-2-Fluor-D-deoxy-glucose uptake (18F-FDG) in tumor cells by various concentrations of FDG carrier or D-glucose in an experimental model using tissue culture and positron emission tomography (PET). Glioblastoma cells in culture were incubated with 18F-FDG with and without added carrier or in presence of glucose concentrations in the range from 0-5 mmol/L. Cellular uptake of 18F-FDG was measured after 20 min. of incubation in PBS-buffer containing different sugar concentrations. The uptake was determined with a PET camera. The similarity of the kinetics of the FDG and glucose uptake are backing the hypothesis that both substrates use the same carrier system. The more intense inhibition of the 18F-uptake by FDG can be explained by the different intracellular metabolism of both substrates. The results explain the clinical experience that there is an optimal 18F-FDG uptake in the patient's tumor when the blood glucose level is as low as possible and the specific activity of 18F-FDG is very high.

[Toxicity and radiation dosage of 2-(18F)-2-desoxy-D-glucose in positron emission tomography]

2-[18F]-FDG, a non-physiological glucose analogue, is the most important positron-emission- tomography (PET) radiopharmaceutical. As an example we refer to the production of 2-[18F]-FDG at the research center in Karlsruhe. 2-[18F]-FDG is synthesized in a "no carrier added" process. It is delivered at a maximal filling volume of 10 ml from a 14.5 ml batch with a batch-to-batch yield fluctuation from 5075 to 50,750 MBq and a specific activity from 1 to 10 GBq/mumol. The residual remaining synthesis reagents like solvents or catalysts have no toxicological relevance. The applicated dose per patient is in a range from 185 to 370 MBq and 1000 times lower than the correlating concentrations of stable FDG which can be regarded harmless in animals. 2-[18F]-FDG does not interfere with normal glucose metabolism. It is taken up by cells and phosphorylated to 2-[18F]-FDG-6-phosphate. The following dephosphorylation step is slow and the labeled compound is retained over several hours within the cells. Non-metabolized 2-[18F]-FDG is excreted rapidly in the urine to an extent of about 16% after 60 min, and 50% after 135 min, respectively. Fluorine-18F decays by emission of 511 KeV gamma photons. The whole body effective dose is reported to be 21 to 27 microSv/MBq. In case of an intravenous injection of 370 MBq this leads to a total dose of 7.8 to 10 mSv. The critical organ is the bladder wall (radiation dose 120 to 170 microSv/MBq or 80 to 100 mrem/mCi). The risk of a radiation induced late malignoma at 10 mSv can be estimated to be 1:2000. The genetical risk as a consequence of FDG-PET diagnostics would be 1:100,000 to 2:100,000 for dominant, and 5 times higher for recessive mutations.

Fluorodeoxyglucose whole-body positron emission tomography in colorectal cancer patients studied in routine daily practice

PURPOSE

To evaluate the routine clinical value of attenuation-corrected whole-body fluorodeoxyglucose positron emission tomography in colorectal cancer, a total of 59 patients who were referred for evaluation of suspected or proven colorectal cancers were studied.

METHODS

Positron emission tomography scans were recorded using a Siemens ECAT Exact 921/47.

RESULTS

Median follow-up after the positron emission tomography study was 11 (mean, 12.3; range, 1-21) months. According to computed tomography, coloscopy, and ultrasound, we recorded eight apparently false-positive results. During later follow-up, however, three of those cases, which were negative with computed tomography, magnetic resonance imaging, sonography, or laparoscopy, turned out to be true-positive instead. In 3 patients, a primary colorectal cancer was suspected; in 26 patients, a recurrence of colorectal cancer was suspected. Eight patients were studied for follow-up after the history of colorectal cancer with no suspicion of recurrence. In 12 patients, the rise of serum tumor marker concentrations was the reason for the positron emission tomography study; 12 patients with known metastatic disease were also included ("restaging"). With regard to the entire patient population, we found an overall sensitivity of 100 percent, a specificity of 67 percent, and positive and negative predictive values of 92 and 100 percent, respectively. Being merely confirmative with respect to tumor recurrence or distant metastases in the majority of patients, positron emission tomography revealed a primary tumor in one patient and confirmed metastatic foci in several patients that had not been delineated by other imaging modalities.

CONCLUSION

A whole-body positron emission tomography scan provides optimum conditions to locate metastatic lesions that might not be seen otherwise. There is a trend showing that positron emission tomography diagnostics as a consequence of early increased tumor markers is a highly sensitive combination, because computed tomography and magnetic resonance imaging were not as sensitive in early recurrences. Positron emission tomography, as performed in daily clinical practice, proved to be a powerful diagnostic tool in our subset of colorectal cancer patients.

[PET scan in general practice for diagnosis of breast carcinoma]

The value of whole body positron emission tomography using 18F-fluoro-deoxy-glucose (18FDG) in primary work-up and follow-up was evaluated retrospectively in 104 patients with primary or metastatic breast cancer. Compared to other imaging methods, FDG-PET sensitivity was superior to sonography, CT or MRT. Another advantage is the possibility of whole body imaging and the earlier detection of lymph node metastasis due to the recognition of functional metabolic changes compared to structural changes found with conventional imaging methods.

Standardized uptake values of fluorine-18 fluorodeoxyglucose: the value of different normalization procedures

While the evident advantages of absolute metabolic rate determinations cannot be equalled by static image analysis of fluorine-18 fluorodeoxyglucose positron emission tomographic (FDG PET) studies, various algorithms for the normalization of static FDG uptake values have been proposed. This study was performed to compare different normalization procedures in terms of dependency on individual patient characteristics. Standardized FDG uptake values (SUVs) were calculated for liver and lung tissue in 126 patients studied with whole-body FDG PET. Uptake values were normalized for total body weight, lean body mass and body surface area. Ranges, means, medians, standard deviations and variation coefficients of these SUV parameters were calculated and their interdependency with total body weight, lean body mass, body surface area, patient height and blood sugar levels was calculated by means of regression analysis. Standardized FDG uptake values normalized for body surface area were clearly superior to SUV parameters normalized for total body weight or lean body mass. Variation and correlation coefficients of body surface area-normalized uptake values were minimal when compared with SUV parameters derived from the other normalization procedures. Normalization for total body weight resulted in uptake values still dependent on body weight and blood sugar levels, while normalization for lean body mass did not eliminate the positive correlation with lean body mass and patient height. It is concluded that normalization of FDG uptake values for body surface area is less dependent on the individual patient characteristics than are FDG uptake values normalized for other parameters, and therefore appears to be preferable for FDG PET studies in oncology.

Fluorine-18 fluorodeoxyglucose positron emission tomography in the follow-up of differentiated thyroid cancer

Whole-body fluorine-18 fluorodeoxyglucose (FDG) positron emission tomography (PET) imaging was performed during the follow-up of 33 patients suffering from differentiated thyroid cancer. Among them there were 26 patients with papillary and seven with follicular tumours. Primary tumour stage (pT) was pT1 in six cases, pT2 in eight cases, pT3 in three cases and pT4 in 14 cases. FDG PET was normal in 18 patients. In three patients a slightly increased metabolism was observed in the thyroid bed, assumed to be related to remnant tissue. In one case local recurrence, in ten cases lymph node metastases (one false-positive, caused by sarcoidosis) and in three cases distant metastases were found with FDG PET. In comparison with whole-body scintigraphy using iodine-131 (WBS) there were a lot of discrepancies in imaging results. Whereas three patients had distant metastases (proven with 131I) and a negative FDG PET, in four cases 131I-negative lymph node metastases were detectable with PET. Even in the patients with concordant "staging", differences between 131I and FDG were observed as to the exact lesion localization. Therefore, a coexistence of 131I-positive/FDG-negative, 131I-negative/FDG-positive and 131I-positive/FDG-positive malignant tissue can be assumed in these patients. A higher correlation of FDG PET was observed with hexakis (2-methoxyisobutylisonitrile) technetium-99m (I) (MIBI) scintigraphy (performed in 20 cases) than with WBS. In highly differentiated tumours 131I scintigraphy had a high sensitivity, whereas in poorly differentiated carcinomas FDG PET was superior. The clinical use of FDG PET can be recommended in all cases of suspected or proven recurrence and/or metastases of differentiated thyroid cancer and is particularly useful in cases with elevated serum thyroglobulin levels and negative WBS.

[Nuclear medicine diagnosis and therapy of distant metastases of pheochromocytoma]

Three foci, suspected of being metastases, were discovered by sonography and computed tomography in a patient twelve years after resection of a phaeochromocytoma and six years after resection of liver metastases. The findings were confirmed by scintigraphy with 131I-metaiodobenzylguanidine (MIBG) and treated with 3700 MBq of MIBG, to be repeated in the near future.

[Does bone metabolic activity determined scintigraphically correlate with calcium and phosphorus measurements by x-ray microanalysis?]

An animal experiment on bone healing of dental implants offered the opportunity of comparing scintigraphic images with histological findings, specially with electron microscope results and analysis of the calcium and phosphorus values in different stages. The clear correspondence between the scintigraphic results and the calcification level of the new bone parts must be seen with reservation because of possible methodical mistakes of the radio-micro-analysis from surfaces and of intraindividual differences.

[Scintigraphy in the diagnosis of intestinal hemorrhage]

13 patients with gastrointestinal hemorrhage underwent a scintigraphic examination to localize the bleeding. In 6 patients Tc-99m labeled red blood cells, in 7 patients Tc-99m sulfur colloid were used. Scintigraphy could demonstrate an intestinal bleeding in ten patients, in whom the conventional examinations, and among them 7 abdominal angiographies had failed. 5 of them had to be operated after that, whereby the scintigraphic findings were verified. One of the three patients with negative scintigraphy had to undergo an operation because of continuous bleeding. The methods of scintigraphy and their indication in gastrointestinal hemorrhage are discussed.

Gelfoam powder embolization of the left gastric artery in treatment of massive small-vessel gastric bleeding

Gelfoam (gelatin foam) powder was used for embolization therapy of massive gastric bleeding from small vessels in 14 patients with severe underlying medical problems. Bleeding was controlled in 10 patients with lesions localized in areas supplied by the embolized left gastric artery. In four patients with concurrent lesions in other portions of the stomach, bleeding decreased only (3 patients) or did not respond to embolization (1 patient). Complications developed in 2 patients with compromised vascular supply of the stomach: superficial ischemic ulcers that healed, and a large ulcer that perforated and required surgery. Microscopic studies demonstrated Gelfoam powder penetration mostly into vessels 100 to 200 microns in diameter and only occasionally into smaller vessels 50 to 60 microns, with occlusion of approximately 10 to 15% of the vasculature. It is concluded that Gelfoam fragments are the primary embolic material to be used for occlusion of the left gastric artery. Use of Gelfoam powder should be limited to occasional patients who have only little chance of responding to Gelfoam fragment embolization. Potential candidates for Gelfoam powder embolization include patients with major coagulopathies and/or uremia who massively hemorrhage from small-vessel lesions localized in upper portions of the stomach, exhibit significant mucosal hypervascularity, and do not respond to selective vasopressin treatment. An uncompromised vascular supply of the stomach is a precondition of a safe left gastric artery embolization.