Assistive AI in Lung Cancer Screening: A Retrospective Multinational Study in the United States and Japan

Purpose To evaluate the impact of an artificial intelligence (AI) assistant for lung cancer screening on multinational clinical workflows. Materials and Methods An AI assistant for lung cancer screening was evaluated on two retrospective randomized multireader multicase studies where 627 (141 cancer-positive cases) low-dose chest CT cases were each read twice (with and without AI assistance) by experienced thoracic radiologists (six U.S.-based or six Japan-based radiologists), resulting in a total of 7524 interpretations. Positive cases were defined as those within 2 years before a pathology-confirmed lung cancer diagnosis. Negative cases were defined as those without any subsequent cancer diagnosis for at least 2 years and were enriched for a spectrum of diverse nodules. The studies measured the readers' level of suspicion (on a 0-100 scale), country-specific screening system scoring categories, and management recommendations. Evaluation metrics included the area under the receiver operating characteristic curve (AUC) for level of suspicion and sensitivity and specificity of recall recommendations. Results With AI assistance, the radiologists' AUC increased by 0.023 (0.70 to 0.72; P = .02) for the U.S. study and by 0.023 (0.93 to 0.96; P = .18) for the Japan study. Scoring system specificity for actionable findings increased 5.5% (57% to 63%; P < .001) for the U.S. study and 6.7% (23% to 30%; P < .001) for the Japan study. There was no evidence of a difference in corresponding sensitivity between unassisted and AI-assisted reads for the U.S. (67.3% to 67.5%; P = .88) and Japan (98% to 100%; P > .99) studies. Corresponding stand-alone AI AUC system performance was 0.75 (95% CI: 0.70, 0.81) and 0.88 (95% CI: 0.78, 0.97) for the U.S.- and Japan-based datasets, respectively. Conclusion The concurrent AI interface improved lung cancer screening specificity in both U.S.- and Japan-based reader studies, meriting further study in additional international screening environments. Keywords: Assistive Artificial Intelligence, Lung Cancer Screening, CT Supplemental material is available for this article. Published under a CC BY 4.0 license.

An intentional approach to managing bias in general purpose embedding models

Advances in machine learning for health care have brought concerns about bias from the research community; specifically, the introduction, perpetuation, or exacerbation of care disparities. Reinforcing these concerns is the finding that medical images often reveal signals about sensitive attributes in ways that are hard to pinpoint by both algorithms and people. This finding raises a question about how to best design general purpose pretrained embeddings (GPPEs, defined as embeddings meant to support a broad array of use cases) for building downstream models that are free from particular types of bias. The downstream model should be carefully evaluated for bias, and audited and improved as appropriate. However, in our view, well intentioned attempts to prevent the upstream components-GPPEs-from learning sensitive attributes can have unintended consequences on the downstream models. Despite producing a veneer of technical neutrality, the resultant end-to-end system might still be biased or poorly performing. We present reasons, by building on previously published data, to support the reasoning that GPPEs should ideally contain as much information as the original data contain, and highlight the perils of trying to remove sensitive attributes from a GPPE. We also emphasise that downstream prediction models trained for specific tasks and settings, whether developed using GPPEs or not, should be carefully designed and evaluated to avoid bias that makes models vulnerable to issues such as distributional shift. These evaluations should be done by a diverse team, including social scientists, on a diverse cohort representing the full breadth of the patient population for which the final model is intended.

Deep Learning Detection of Active Pulmonary Tuberculosis at Chest Radiography Matched the Clinical Performance of Radiologists

Background The World Health Organization (WHO) recommends chest radiography to facilitate tuberculosis (TB) screening. However, chest radiograph interpretation expertise remains limited in many regions. Purpose To develop a deep learning system (DLS) to detect active pulmonary TB on chest radiographs and compare its performance to that of radiologists. Materials and Methods A DLS was trained and tested using retrospective chest radiographs (acquired between 1996 and 2020) from 10 countries. To improve generalization, large-scale chest radiograph pretraining, attention pooling, and semisupervised learning ("noisy-student") were incorporated. The DLS was evaluated in a four-country test set (China, India, the United States, and Zambia) and in a mining population in South Africa, with positive TB confirmed with microbiological tests or nucleic acid amplification testing (NAAT). The performance of the DLS was compared with that of 14 radiologists. The authors studied the efficacy of the DLS compared with that of nine radiologists using the Obuchowski-Rockette-Hillis procedure. Given WHO targets of 90% sensitivity and 70% specificity, the operating point of the DLS (0.45) was prespecified to favor sensitivity. Results A total of 165 754 images in 22 284 subjects (mean age, 45 years; 21% female) were used for model development and testing. In the four-country test set (1236 subjects, 17% with active TB), the receiver operating characteristic (ROC) curve of the DLS was higher than those for all nine India-based radiologists, with an area under the ROC curve of 0.89 (95% CI: 0.87, 0.91). Compared with these radiologists, at the prespecified operating point, the DLS sensitivity was higher (88% vs 75%, P < .001) and specificity was noninferior (79% vs 84%, P = .004). Trends were similar within other patient subgroups, in the South Africa data set, and across various TB-specific chest radiograph findings. In simulations, the use of the DLS to identify likely TB-positive chest radiographs for NAAT confirmation reduced the cost by 40%-80% per TB-positive patient detected. Conclusion A deep learning method was found to be noninferior to radiologists for the determination of active tuberculosis on digital chest radiographs. © RSNA, 2022 Online supplemental material is available for this article. See also the editorial by van Ginneken in this issue.

Congo red dot test in the early prediction and diagnosis of pre-eclampsia in a tertiary health care centre in India

OBJECTIVES

To demonstrate the use of urine congophilia quantification in the prediction and diagnosis of pre-eclampsia using Congo red dot test.

STUDY DESIGN

A prospective cohort study in 378 consecutive pregnant women was conducted. All eligible, consenting women of gestational age between 10 and 34 weeks were enrolled in the study. The presence of urinary misfolded proteins was screened by a simple dot test technique on unsupported nitrocellulose membrane using Congo red dye.

RESULTS

The urinary congophilia was increased in urine from women with pre eclampsia compared to healthy pregnant controls. The mean CRR value of pre eclamptic pregnant women (35.2 ± 9.4%) was five times higher than that of mean CRR value of normotensive pregnant women (6.9 ± 4.7%). The mean gestational age at which Congo red test showed positive was 26.95 ± 2.90 weeks and the time taken from CRD positive to development of PE was 4.92 ± 2.54 weeks of gestation.

CONCLUSIONS

In our study, the CRD test was not only effective in predicting pre-eclampsia but was also useful in differentiating between pre-eclampsia and other forms of hypertension, as well as early onset and late onset pre-eclampsia, with positive predictive value of 80.36% and negative predictive value of 92.86.

Characterization of Rubia cordifolia L. root extract and its evaluation of cardioprotective effect in Wistar rat model

OBJECTIVES:

Rubia cordifolia L. (RC) is a well-known and highly valuable medicinal plant in the Ayurvedic system. The present study involves evaluating antioxidant and cardioprotective property of RC root extract.

MATERIALS AND METHODS:

The characterization of RC root extract was carried out using standard phytochemical and biochemical analysis. The functional groups were analyzed by Fourier transform infrared (FTIR) spectroscopy and phytotherapeutic compounds were identified using high-resolution mass spectrometry (HR-MS). Cardioprotective activity of RC root extract was investigated against cyclophosphamide (CP; 100 mg/kg, i.p)-induced cardiotoxicity in male albino Wistar rats. RC (100, 200, and 400 mg/kg, p.o) or silymarin (100 mg/kg, p.o) was administered immediately after CP on the 1^st day and the next consecutive 10 days. Biochemical and histopathological analysis was performed to observe the cardioprotective effects of RC root extract.

RESULTS:

Phytochemical analysis revealed the presence of secondary metabolites that include alkaloids, flavonoids, saponins, and anthraquinones in RC root extract. FTIR analysis revealed the presence of several functional groups. Based on HR-MS analysis, eight major phytotherapeutic compounds were identified in methanol root extract of RC. Biochemical analysis in CP-induced rat model administered with RC extract revealed significantly enhanced levels of antioxidant markers such as superoxide dismutase, catalase, and glutathione S-transferase. Histopathological study showed that the rat model treated with the root extract had reduced the cardiac injury.

CONCLUSION:

Our results have shown that the RC extract contains various antioxidant compounds with cardioprotective effect. Treatment with RC root extract could significantly protect CP-induced rats from cardiac tissue injury by restoring the antioxidant markers.

Molecular analysis of PRRT2 gene in a case of paroxysmal kinesigenic dyskinesia patient

Paroxysmal kinesigenic dyskinesia (PKD) is an abnormal involuntary movement that is episodic or intermittent, with sudden onset, and the attacks are induced by sudden movement. Mutations in proline-rich transmembrane protein 2 (PRRT2) gene have been implicated in the cause of this disorder. This study presents a case of PKD on the basis of clinical findings supported and evidences obtained through a mutational analysis. Sequencing of all the exons of PRRT2 gene revealed a frameshift mutation (p.R217Pfs*8) in exon 2 and a novel transition mutation (c.244C > T) in 5′-untranslated region (UTR). Though mutations in PRRT2 gene are well-established in PKD, this study for the first time presents a novel transition mutation in the exon 2 region.

Modulation of cellular radiation responses by 2-deoxy-D-glucose and other glycolytic inhibitors: implications for cancer therapy

BACKGROUND

2-Deoxy-D-glucose (2-DG), a glycolytic inhibitor, was observed earlier to increase DNA, chromosomal, and cellular damage in tumor cells, by inhibiting energy-dependent repair processes. Lonidamine (LND) selectively inhibits glycolysis in cancer cells. It damages the condensed mitochondria in these cells, impairing thereby the activity of hexokinase (predominantly attached to the outer mitochondrial membranes). It inhibits repair of radiation-induced potentially lethal cellular damage in HeLa and Chinese hamster (HA-1) cells. However, other than a preliminary study on human glioma (BMG-1) cells in this laboratory, the effects of LND on radiation-induced cytogenetic damage have not been reported earlier.

AIMS

These studies were carried out to investigate the effects of LND and 2-DG on cell proliferation, viability, and radiation response in the same human glioma cell line, under identical conditions. The respective drug concentrations were selected on the basis of earlier studies.

MATERIALS AND METHODS

Human glioma (U373MG) cells were grown in the presence of LND or 2-DG for 2 days. Proliferation response and viability of U373MG human glioma cells were studied by cell counts and uptake of trypan blue dye. Radiosensitization (increase in micronuclei formation) was studied after short-term (4 h postirradiation) drug treatments.

OBSERVATIONS

Both the drugs (1) inhibited proliferation response in a concentration-dependent manner; (2) did not induce micronuclei formation in the unirradiated cells; and (3) significantly increased radiation-induced micronuclei formation at nontoxic concentrations.

CONCLUSIONS

These data suggest that the short-term presence of either lonidamine or 2-DG-at clinically relevant and nontoxic concentrations-could increase the treatment response of malignant gliomas at optimum radiation doses, reducing thereby the side effects of radiotherapy.

Optimizing radiotherapy of brain tumours by a combination of temozolomide & lonidamine

BACKGROUND & OBJECTIVE

Temozolomide (TMZ), a second generation alkylating drug, an effective cytotoxic agent as well as radiosensitizer for malignant brain tumours, has side effects like myelosuppression. Lonidamine (LND) increases the effectiveness of several experimental multiple chemotherapy protocols, without increasing bone marrow toxicities and is effective in brain tumour patients. The objective of the present studies was to investigate whether combining clinically relevant doses of LND and TMZ could increase the proliferation and radiation response of malignant human brain tumour cells in vitro.

METHODS

A malignant human glioma (U373MG) cell line was used in these studies. TMZ (20, 40 or 60 microM) or LND (100, 150 or 200 microM), or the combination of both (20 and 100 microM, respectively) in 0.1 per cent dimethyl sulphoxide (DMSO) were added three days after setting up cultures, in six well plates (5 x 10(4) cells/ well). The effects of continuous treatment for two days on proliferation response and cytotoxicity were studied after trypsinization; by cell counts and the uptake of trypan blue dye (0.5%). For the study of radiation (60Co-Gamma-rays, 2 Gy) response, drugs were removed 4 h after irradiation and cultures were grown further in drug free, normal growth medium for another 20 h or 44 h.

RESULTS

Continuous presence of TMZ or LND for two days significantly inhibited cell proliferation in a concentration dependent manner. The frequencies of non viable cells increased significantly only at higher concentrations of LND. Combination of 20 microM TMZ with 100 microM LND had additive effects on proliferation response, without affecting cell viability. Short-term drug treatments without irradiation did not induce micronuclei formation. Cell proliferation and viability were also not affected. However, post-irradiation presence of either of these drugs for 4 h significantly reduced the proliferation response, 24 and 48 h after treatments. It was further inhibited by the combination treatment. On the contrary, radiation induced micronuclei formation was enhanced by either of the drugs; which was significantly increased by the combined treatment, 24 h as well as 48 h after irradiation. No effects on cell viability were observed, immediately after these treatments as well as at later time points.

INTERPRETATION & CONCLUSION

Our findings showed that combination of TMZ and LND at clinically achievable, low plasma concentrations could inhibit tumour growth, and lonidamine could reduce the dose of temozolomide required for radiosensitization of brain tumours.