Standalone deep learning versus experts for diagnosis lung cancer on chest computed tomography: a systematic review.

PURPOSE: To compare the diagnostic performance of standalone deep learning (DL) algorithms and human experts in lung cancer detection on chest computed tomography (CT) scans. MATERIALS AND METHODS: This study searched for studies on PubMed, Embase, and Web of Science from their inception until November 2023. We focused on adult lung cancer patients and compared the efficacy of DL algorithms and expert radiologists in disease diagnosis on CT scans. Quality assessment was performed using QUADAS-2, QUADAS-C, and CLAIM. Bivariate random-effects and subgroup analyses were performed for tasks (malignancy classification vs invasiveness classification), imaging modalities (CT vs low-dose CT [LDCT] vs high-resolution CT), study region, software used, and publication year. RESULTS: We included 20 studies on various aspects of lung cancer diagnosis on CT scans. Quantitatively, DL algorithms exhibited superior sensitivity (82%) and specificity (75%) compared to human experts (sensitivity 81%, specificity 69%). However, the difference in specificity was statistically significant, whereas the difference in sensitivity was not statistically significant. The DL algorithms' performance varied across different imaging modalities and tasks, demonstrating the need for tailored optimization of DL algorithms. Notably, DL algorithms matched experts in sensitivity on standard CT, surpassing them in specificity, but showed higher sensitivity with lower specificity on LDCT scans. CONCLUSION: DL algorithms demonstrated improved accuracy over human readers in malignancy and invasiveness classification on CT scans. However, their performance varies by imaging modality, underlining the importance of continued research to fully assess DL algorithms' diagnostic effectiveness in lung cancer. CLINICAL RELEVANCE STATEMENT: DL algorithms have the potential to refine lung cancer diagnosis on CT, matching human sensitivity and surpassing in specificity. These findings call for further DL optimization across imaging modalities, aiming to advance clinical diagnostics and patient outcomes. KEY POINTS: Lung cancer diagnosis by CT is challenging and can be improved with AI integration. DL shows higher accuracy in lung cancer detection on CT than human experts. Enhanced DL accuracy could lead to improved lung cancer diagnosis and outcomes.

Successful rechallenge of alectinib after remission of severe alectinib-induced interstitial lung disease.

INTRODUCTION: Even though alectinib is a potent second-generation ALK inhibitor with a favorable safety profile, alectinib-induced interstitial lung disease (ILD) could be fatal. There are case reports described successful alectinib rechallenge in mild ILD. However, the feasibility and safety of rechallenge in severe cases remains to be elucidated. CASE REPORT: A 76-year-old female was a case of stage IV lung adenocarcinoma harboring ALK rearrangement. Respiratory failure following severe ILD developed one month after alectinib administration. She received mechanical ventilation in intensive care uint. ILD subsided gradually after methylprednisolone pulse therapy and discontinuation of alectinib.Management and outcome: After the recovery from ILD, the patient attempted a re-escalation of alectinib from a lower dose under close clinical and radiological monitoring. No ILD happened even after 480 days of alectinib rechallenge. DISCUSSION: Given that the ALK inhibitors are the treatment of choice for advanced lung cancer patients with ALK rearrangement. Our report demonstrated the potential feasibility of alectinib re-use in cases of severe druginduced ILD.

Effects of rapid ascent on the heart rate variability of individuals with and without acute mountain sickness.

PURPOSE: Through time- and frequency-domain analysis, we compared the effects of acute hypobaric hypoxia on the changes in heart rate variability (HRV) following night sleeping and morning awakening in individuals with and without acute mountain sickness (AMS). METHOD: Thirty-nine nonacclimatised healthy individuals were transported by bus from sea level to 3150 m within 3 h. Short-term HRV was measured two times a day-before sleeping (BS) and after awakening (AA)- at 3 days before ascent (T0), two consecutive nights at 3150 m (T1 and T2), and 2 days after descent (T3). AMS was diagnosed using the self-reported Lake Louise score questionnaire. RESULT: AMS developed in 19 of 39 participants (48.7%). At sea level, individuals had higher HRV at AA than at BS, and the trend of increased HRV at AA remained unchanged at high altitude, irrespective of AMS. At T1 BS, low-frequency power in normalised unit was significantly lower in participants with AMS than in those without AMS. Compared with those at T1 BS, the square root of the mean squared differences of successive normal-normal (NN) intervals, the number of interval differences of successive NN intervals more than 50 ms (NN50), and the proportion derived by dividing NN50 by the total number of NN intervals at T1 AA significantly increased in participants without AMS but nonsignificantly decreased in those with AMS. CONCLUSION: After rapid ascent, individuals with AMS did not demonstrate sympathetic hyperactivity but did exhibit withdrawal of cardiac vagal modulation in the morning following the first night's sleep.

Impact of rapid ascent to high altitude on sleep.

PURPOSE: Sleep disturbance at high altitude is common in climbers. In this study, we intended to evaluate the effect of rapid ascent on sleep architecture using polysomnography (PSG) and to compare the differences between subjects with and without acute mountain sickness (AMS). METHODS: The study included 40 non-acclimatized healthy subjects completing PSG at four time points, 3 days before the ascent (T0), two successive nights at 3150 m (T1 and T2), and 2 days after the descent (T3). All subjects were transported by bus from 555 to 3150 m within 3 h. AMS was diagnosed using self-reported questionnaire of Lake Louise score. RESULTS: Twenty of 40 (50%) subjects developed AMS. At high altitude, awakening percentages increased in AMS group but changed insignificantly in non-AMS group. Arousal index and apnea/hypopnea index (AHI) increased irrespective of AMS. The increases of AHI were more evident in non-AMS group than in AMS group. Compared to subjects without AMS, those with AMS had significantly lower sleep efficiency, lower central apnea index, and longer latencies to sleep and rapid eye movement (REM) sleep at T1 and lower REM sleep percentages at T1 and T2. Subjects with older age and lower minimum arterial oxygen saturation during sleep at sea level were prone to develop AMS. CONCLUSIONS: Higher AHI did not cause more frequent awakenings and arousals at high altitude. Central sleep apneas were observed in non-AMS but not in AMS group. Subjects unacclimatized to acute hypobaric hypoxia might have delayed and less REM sleep.

Real-world osimertinib pretreatment experience in patients with epidermal growth factor receptor T790M mutation-positive locally advanced or metastatic non-small cell lung cancer.

Osimertinib has demonstrated efficacy in patients with epidermal growth factor receptor (EGFR) T790M-positive non-small cell lung cancer (NSCLC) in clinical trials. However, real-world data on its effectiveness remain scarce. Taiwanese patients with T790M-positive locally advanced or metastatic NSCLC and progressive disease following treatment with at least one EGFR tyrosine kinase inhibitor (TKI) were enrolled from the osimertinib early access program. Of the 419 patients (mean age, 63 years; female, 67%), 53% were heavily pretreated (≥ third-line [3L]), making osimertinib a fourth-line (4L) intervention. The median progression-free survival (PFS) was 10.5 months (95% confidence interval [CI]: 8.95-11.41); the 18-month PFS rate was 26.5%. The median overall survival (OS) was 19.0 months (95% CI: 16.30-20.95); the 24-month OS rate was 40.9%. The objective response rate was 32.46%, and the disease control rate was 86.38%. The median time to treatment discontinuation of osimertinib monotherapy was 11.9 months (95% CI: 10.49-13.11). Subgroup analyses of median PFS and OS in the chemotherapy combination group vs. the osimertinib monotherapy group yielded no difference. Central nervous system (CNS) metastasis, number of prior lines of therapy, and types of initial EGFR-TKIs did not significantly impact outcomes. The median PFS values were 9.0 (95% CI: 5.18-11.34) and 10.9 (95% CI: 9.18-11.90) months with and without CNS metastasis, respectively, and 10.8 (95% CI: 8.59-12.69), 13.6 (95% CI: 10.89-16.3), and 9.2 (95% CI: 7.8-10.62) months for second-line (2L), 3L, and ≥4L therapy, respectively. In patients who received osimertinib as 2L therapy, the median PFS values in response to prior afatinib, erlotinib and gefitinib treatment were 11.2 (95% CI: 4.85-4.79), 10.5 (95% CI: 8.59-20.26) and 8.7 (95% CI: 7.21-16.79) months, respectively. Overall, real-world data from Taiwan support the clinical benefits of osimertinib in EGFR T790M -positive NSCLC.

Clinical utility of immunohistochemical subtyping in patients with small cell lung cancer.

OBJECTIVES: Molecular subtyping of small cell lung cancer (SCLC) tumors based on the expression of four transcription factors (ASCL1, NEUROD1, POU2F3, and YAP1) using immunohistochemical (IHC) staining has recently emerged as a proposed approach. This study was aimed to examine this subtyping method in Asian patients with SCLC and investigate its correlation with treatment efficacy. MATERIALS AND METHODS: Seventy-two tumor samples from patients with SCLC, including de novo cases and those transformed from EGFR-mutant tumors, were analyzed. IHC staining was used to measure the expression of the four transcription factors and conventional SCLC markers. Subtypes were defined based on relative expression levels. The treatment response and outcome of patients receiving immune checkpoint inhibitors and chemotherapy were also reviewed. RESULTS: ASCL1 was the most common subtype, observed in 55.2 % of the samples, followed by NEUROD1 (26.9 %) and POU2F3 (9 %). No tumor exhibited predominant YAP1 positivity, while 41.8 % of the samples demonstrated positivity for two subtype markers. Approximately 50 % of the patients experienced a subtype switch after disease progression. Patients with the ASCL1/NEUROD1 (SCLC-A/N) subtype had similar progression-free survival (PFS) compared to non-SCLC-A/N patients after treatment with immune checkpoint inhibitors plus chemotherapy. Transformed SCLC patients had significantly worse PFS than de novo SCLC patients after chemoimmunotherapy. (2.1 vs. 5.4 months, P = 0.023) CONCLUSIONS: This study revealed the challenges associated with using IHC alone for molecular subtyping, highlighting the frequent co-expression of subtypes and temporal changes following treatment. Further research is warranted to explore the prognostic and therapeutic implications of IHC subtyping in patients with SCLC.

Statin-induced lung injury: diagnostic clue and outcome.

BACKGROUND: Statin-induced lung injury (SILI) is an uncommon but serious complication of statins. The clinical features and outcome of patients with SILI vary widely. Clinical data relevant to diagnosis and outcome of patients with SILI were investigated in this study. METHOD: Four cases of SILI diagnosed at our institute and 12 cases reported in the English literature from 1995 to 2010 were studied. The patients were further divided into favourable and unfavourable outcome groups and compared. RESULTS: Compared with the 12 previously reported cases, fever (p=0.008) and consolidation (p=0.027) were more common and duration of statin treatment was significantly shorter (p=0.030) in our patients. Foamy alveolar macrophages in bronchoalveolar lavage fluid (BALF) were found in our four patients. Patients with cough (p=0.024), fever (p=0.026) and alveolar infiltrates (p=0.036), especially ground-glass opacity (GGO) (p=0.001) shown on thoracic high-resolution CT (HRCT), had a favourable outcome. Conversely, those with fibrosis shown on HRCT (p=0.008) had an unfavourable outcome. Stepwise logistic regression analysis demonstrated that cough (p=0.011), fever (p=0.005), and alveolar infiltrates (p=0.017), GGO (p<0.001) and fibrosis (p=0.002) shown on thoracic HRCT were independent factors affecting the outcome of SILI. CONCLUSIONS: For patients with SILI, pulmonary phospholipidosis, as shown by foamy alveolar macrophages in BALF, may be valuable in diagnosis, and clinical symptoms and thoracic HRCT findings are of value in predicting the outcome.

Temporal changes in biomarkers in individuals with and without acute mountain sickness following rapid ascent.

BACKGROUND: Field studies have reported conflicting results regarding changes in biomarkers at high altitude. This study measured temporal changes in biomarkers and compared the differences between individuals with and without acute mountain sickness (AMS). MATERIALS AND METHODS: This study included 34 nonacclimatized healthy participants. Ten-milliliters of blood were collected at four time points: 3 days before ascent (T0), on two successive nights at 3150 m (T1 and T2), and 2 days after descent (T3). Participants were transported by bus from 555 m to 3150 m within 3 hours. AMS was diagnosed using the self-reported Lake Louise Scoring (LLS) questionnaire. RESULTS: Compared with T0, significant increases in E-selectin and decreases in vascular endothelial growth factor (VEGF) levels were observed at high altitude. Significantly increased C-reactive protein (CRP), monocyte chemoattractant protein-1 (MCP-1), and S100 calcium-binding protein B (S100B) levels were observed at T2, and significantly decreased vascular cell adhesion molecule-1 (VCAM-1) levels were observed at T3. Eighteen (53%) participants developed AMS. Changes in E-selectin, CRP, MCP-1, and S100B levels were independent of AMS. Relative to individuals without AMS, those with AMS had significantly higher atrial natriuretic peptide (ANP) and VCAM-1 levels and lower plasminogen activator inhibitor-1 (PAI-1) levels at T1 and higher brain natriuretic peptide and lower VEGF and PAI-1 levels at T3. LLSs were positively correlated with ANP and VCAM-1 levels and negatively correlated with PAI-1 levels measured at T1. CONCLUSIONS: After acute ascent, individuals with and without AMS exhibited different trends in biomarkers associated with endothelial cell activation and natriuretic peptides.

Endobronchial ultrasound-guided transbronchial needle aspiration for the diagnosis of pulmonary sarcoidosis: A 9-year experience at a single center.

BACKGROUND: Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is valuable for diagnosing pulmonary sarcoidosis. We aimed to evaluate the diagnostic yield of EBUS-TBNA and cytology in sarcoidosis during the first 9 years at our institution. METHODS: Patients who underwent EBUS-TBNA for suspected sarcoidosis between January 2011 and November 2019 were identified retrospectively. EBUS-TBNA was performed with rapid on-site cytological evaluation of the samples. The final diagnosis was based on the pathology and/or cytology results, radiologic features, and clinical follow-up findings. The yield rate was analyzed annually. RESULTS: Eighty patients underwent 83 EBUS-TBNA procedures for suspected sarcoidosis. In total, 136 lymph nodes were sampled. The mean number of lymph node stations sampled was 2.0 ± 0.6; the mean number of needle passes per lymph node was 3.5 ± 0.8. Sixty-five patients were diagnosed with sarcoidosis, with a total of 68 procedures. Nonnecrotizing granulomatous inflammation was detected in the EBUS-TBNA samples from 49/68 procedures (yield rate: 72.1%). Of 19 patients with sarcoidosis who did not obtain a pathological diagnosis with EBUS-TBNA, epithelioid cells and/or multinuclear giant cells suggestive of granulomatous inflammation were detected in five. The sensitivity, specificity, positive predictive value, and negative predictive value (NPV) for pathological diagnosis of sarcoidosis using EBUS-TBNA were 72.1%, 100%, 100%, and 24.0%, respectively. On using cytology, the sensitivity and NPV increased to 79.4% and 26.3%, respectively. The yield rate did not increase until 2016. CONCLUSION: EBUS-TBNA is useful for diagnosing sarcoidosis. Cytology resulted in an additional yield rate of 7.3%, which improved as the number of cases increased.

Application of Radiomics in Prognosing Lung Cancer Treated with Epidermal Growth Factor Receptor Tyrosine Kinase Inhibitors: A Systematic Review and Meta-Analysis.

In the context of non-small cell lung cancer (NSCLC) patients treated with EGFR tyrosine kinase inhibitors (TKIs), this research evaluated the prognostic value of CT-based radiomics. A comprehensive systematic review and meta-analysis of studies up to April 2023, which included 3111 patients, was conducted. We utilized the Quality in Prognosis Studies (QUIPS) tool and radiomics quality scoring (RQS) system to assess the quality of the included studies. Our analysis revealed a pooled hazard ratio for progression-free survival of 2.80 (95% confidence interval: 1.87-4.19), suggesting that patients with certain radiomics features had a significantly higher risk of disease progression. Additionally, we calculated the pooled Harrell's concordance index and area under the curve (AUC) values of 0.71 and 0.73, respectively, indicating good predictive performance of radiomics. Despite these promising results, further studies with consistent and robust protocols are needed to confirm the prognostic role of radiomics in NSCLC.

A radiomics-based deep learning approach to predict progression free-survival after tyrosine kinase inhibitor therapy in non-small cell lung cancer.

BACKGROUND: The epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) are a first-line therapy for non-small cell lung cancer (NSCLC) with EGFR mutations. Approximately half of the patients with EGFR-mutated NSCLC are treated with EGFR-TKIs and develop disease progression within 1 year. Therefore, the early prediction of tumor progression in patients who receive EGFR-TKIs can facilitate patient management and development of treatment strategies. We proposed a deep learning approach based on both quantitative computed tomography (CT) characteristics and clinical data to predict progression-free survival (PFS) in patients with advanced NSCLC after EGFR-TKI treatment. METHODS: A total of 593 radiomic features were extracted from pretreatment chest CT images. The DeepSurv models for the progression risk stratification of EGFR-TKI treatment were proposed based on CT radiomic and clinical features from 270 stage IIIB-IV EGFR-mutant NSCLC patients. Time-dependent PFS predictions at 3, 12, 18, and 24 months and estimated personalized PFS curves were calculated using the DeepSurv models. RESULTS: The model combining clinical and radiomic features demonstrated better prediction performance than the clinical model. The model achieving areas under the curve of 0.76, 0.77, 0.76, and 0.86 can predict PFS at 3, 12, 18, and 24 months, respectively. The personalized PFS curves showed significant differences (p < 0.003) between groups with good (PFS > median) and poor (PFS < median) tumor control. CONCLUSIONS: The DeepSurv models provided reliable multi-time-point PFS predictions for EGFR-TKI treatment. The personalized PFS curves can help make accurate and individualized predictions of tumor progression. The proposed deep learning approach holds promise for improving the pre-TKI personalized management of patients with EGFR-mutated NSCLC.

Artificial Intelligence Assisted Computational Tomographic Detection of Lung Nodules for Prognostic Cancer Examination: A Large-Scale Clinical Trial.

Low-dose computed tomography (LDCT) has emerged as a standard method for detecting early-stage lung cancer. However, the tedious computer tomography (CT) slide reading, patient-by-patient check, and lack of standard criteria to determine the vague but possible nodule leads to variable outcomes of CT slide interpretation. To determine the artificial intelligence (AI)-assisted CT examination, AI algorithm-assisted CT screening was embedded in the hospital picture archiving and communication system, and a 200 person-scaled clinical trial was conducted at two medical centers. With AI algorithm-assisted CT screening, the sensitivity of detecting nodules sized 4−5 mm, 6~10 mm, 11~20 mm, and >20 mm increased by 41%, 11.2%, 10.3%, and 18.7%, respectively. Remarkably, the overall sensitivity of detecting varied nodules increased by 20.7% from 67.7% to 88.4%. Furthermore, the sensitivity increased by 18.5% from 72.5% to 91% for detecting ground glass nodules (GGN), which is challenging for radiologists and physicians. The free-response operating characteristic (FROC) AI score was ≥0.4, and the AI algorithm standalone CT screening sensitivity reached >95% with an area under the localization receiver operating characteristic curve (LROC-AUC) of >0.88. Our study demonstrates that AI algorithm-embedded CT screening significantly ameliorates tedious LDCT practices for doctors.

Evaluating the Potential of Delta Radiomics for Assessing Tyrosine Kinase Inhibitor Treatment Response in Non-Small Cell Lung Cancer Patients.

Our study aimed to harness the power of CT scans, observed over time, in predicting how lung adenocarcinoma patients might respond to a treatment known as EGFR-TKI. Analyzing scans from 322 advanced stage lung cancer patients, we identified distinct image-based patterns. By integrating these patterns with comprehensive clinical information, such as gene mutations and treatment regimens, our predictive capabilities were significantly enhanced. Interestingly, the precision of these predictions, particularly related to radiomics features, diminished when data from various centers were combined, suggesting that the approach requires standardization across facilities. This novel method offers a potential pathway to anticipate disease progression in lung adenocarcinoma patients treated with EGFR-TKI, laying the groundwork for more personalized treatments. To further validate this approach, extensive studies involving a larger cohort are pivotal.

Characteristics, treatment patterns, and clinical outcomes in patients with advanced non-small cell lung cancer harboring EGFR exon 20 insertions.

PURPOSE: Epidermal growth factor receptor (EGFR) exon 20 insertions (ex20ins) are associated with poor prognosis and resistance to traditional therapies in patients with non-small cell lung cancer (NSCLC). We aimed to elucidate the characteristics and treatment patterns to improve outcomes among this population in Taiwan. METHODS: Patients with advanced or recurrent NSCLC harboring EGFR ex20ins from 2011 to 2021 were reviewed. The treatment groups were classified as platinum-based chemotherapy (PtC), EGFR tyrosine kinase inhibitor (TKI), and others. The response to therapy, objective response rate (ORR), disease control rate (DCR), overall survival (OS), progression-free survival (PFS), and factors associated with survival were analyzed. RESULTS: Among the 71 patients, most were never-smoking males with stage IVB adenocarcinoma. The most common first-line (1L) regimen was PtC, followed by TKI. The most common second-line (2L) regimen was TKI. The median PFS of 1L treatment was 5.03 months, and the median OS was 18.43 months. Compared with that of TKI, 1L PtC use was associated with a higher ORR (26.3% vs. 9.1%) and DCR (60.5% vs. 18.2%) and a longer PFS (5.37 vs. 3.13 months, p = 0.044). PFS was also significantly longer in the 2L PtC group than in the 2L TKI group (4.73 vs. 2.25 months, p = 0.047). No patients receiving an immune checkpoint inhibitor-based regimen exhibited a therapeutic response. CONCLUSION: This study demonstrated the heterogeneous clinical characteristics and treatment pattern of NSCLC patients with EGFR ex20ins, underscoring the need for more effective therapies for this distinct molecular subtype.

Personalized cancer avatars for patients with thymic malignancies: A pilot study with circulating tumor cell-derived organoids.

BACKGROUND: Systemic therapy is the primary treatment for advanced thymic malignancies. However, there is an urgent need to improve clinical outcome. Personalized treatment based on predictive biomarkers is a potential approach to address this requirement. In this study, we aimed to show the correlation between drug sensitivity tests on CTCs-derived organoids and clinical response in patients with thymic malignancies. This approach carries the potential to create personalized cancer avatars and improve treatment outcome for patients. METHODS: We previously reported potential treatment outcome prediction with patient-derived organoids (cancer avatars) in patients with pancreatic ductal adenocarcinoma. To further investigate the feasibility of this approach in advanced thymic malignancies, we conducted a study in which 12 patients were enrolled and 21 liquid biopsies were performed. RESULTS: Cancer avatars were successfully derived in 16 out of 21 samples (success rate 76.2%). We found a sensitivity of 1.0 and specificity of 0.6 for drug sensitivity tests on the cancer avatars, and a two-tailed Fisher's exact test revealed a significant correlation between drug sensitivity tests and clinical responses (p = 0.0275). CONCLUSION: This study supports the potential of circulating tumor cell-derived organoids to inform personalized treatment for advanced thymic malignancies. Further validation of this proof of concept finding is ongoing.

Old age and EGFR mutation status in inoperable early-stage non-small cell lung cancer patients receiving stereotactic ablative radiotherapy: A single institute experience of 71 patients in Taiwan.

BACKGROUND: Stereotactic ablative radiotherapy (SABR) is now the standard of care for patients with inoperable early-stage lung cancer. Many of these patients are elderly. EGFR (epidermal growth factor receptor) mutation is also common in the Asian population. METHODS: To evaluate the effects of old age and EGFR mutation on treatment outcomes and toxicity, we reviewed the medical records of 71 consecutive patients with inoperable early-stage non-small cell lung cancer (NSCLC) who received SABR at Taipei Veterans General Hospital between 2015 and 2021. RESULTS: The study revealed that median age, follow-up, Charlson comorbidity index, and ECOG score were 80 years, 2.48 years, 3, and 1, respectively. Of these patients, 37 (52.1%) were 80 years or older, and 50 (70.4%) and 21 (29.6%) had T1 and T2 diseases, respectively. EGFR mutation status was available for 33 (46.5%) patients, of whom 16 (51.5%) had a mutation. The overall survival rates at 1, 3, and 5 years were 97.2, 74.9, and 58.3%, respectively. The local control rate at 1, 3, and 5 years was 97.1, 92.5, and 92.5%, respectively. Using Cox proportional hazards regression we found that male sex was a risk factor for overall survival (p = 0.036, 95% CI: 1.118-26.188). Two patients had grade 2 pneumonitis, but no other grade 2 or higher toxicity was observed. We did not find any significant differences in treatment outcomes or toxicity between patients aged 80 or older and those with EGFR mutations in this cohort. CONCLUSION: These findings indicate that age and EGFR mutation status do not significantly affect the effectiveness or toxicity of SABR for patients with inoperable early-stage NSCLC.

Application of Artificial Intelligence in Lung Cancer.

Lung cancer is the leading cause of malignancy-related mortality worldwide due to its heterogeneous features and diagnosis at a late stage. Artificial intelligence (AI) is good at handling a large volume of computational and repeated labor work and is suitable for assisting doctors in analyzing image-dominant diseases like lung cancer. Scientists have shown long-standing efforts to apply AI in lung cancer screening via CXR and chest CT since the 1960s. Several grand challenges were held to find the best AI model. Currently, the FDA have approved several AI programs in CXR and chest CT reading, which enables AI systems to take part in lung cancer detection. Following the success of AI application in the radiology field, AI was applied to digitalized whole slide imaging (WSI) annotation. Integrating with more information, like demographics and clinical data, the AI systems could play a role in decision-making by classifying EGFR mutations and PD-L1 expression. AI systems also help clinicians to estimate the patient's prognosis by predicting drug response, the tumor recurrence rate after surgery, radiotherapy response, and side effects. Though there are still some obstacles, deploying AI systems in the clinical workflow is vital for the foreseeable future.

Generating High-Resolution CT Slices from Two Image Series Using Deep-Learning-Based Resolution Enhancement Methods.

Medical image super-resolution (SR) has mainly been developed for a single image in the literature. However, there is a growing demand for high-resolution, thin-slice medical images. We hypothesized that fusing the two planes of a computed tomography (CT) study and applying the SR model to the third plane could yield high-quality thin-slice SR images. From the same CT study, we collected axial planes of 1 mm and 5 mm in thickness and coronal planes of 5 mm in thickness. Four SR algorithms were then used for SR reconstruction. Quantitative measurements were performed for image quality testing. We also tested the effects of different regions of interest (ROIs). Based on quantitative comparisons, the image quality obtained when the SR models were applied to the sagittal plane was better than that when applying the models to the other planes. The results were statistically significant according to the Wilcoxon signed-rank test. The overall effect of the enhanced deep residual network (EDSR) model was superior to those of the other three resolution-enhancement methods. A maximal ROI containing minimal blank areas was the most appropriate for quantitative measurements. Fusing two series of thick-slice CT images and applying SR models to the third plane can yield high-resolution thin-slice CT images. EDSR provides superior SR performance across all ROI conditions.

Real-world evidence of the intrinsic limitations of PCR-based EGFR mutation assay in non-small cell lung cancer.

Detection of driver gene mutations is important in advanced NSCLC. The cobas EGFR mutation test is a mutant allele-specific real-time PCR assay with limitation owing to its primer design. Next-generation sequencing-based assay has a higher mutation detection coverage; however, its clinical impact remains unclear. We retrospectively collected the records of stage IV NSCLC patients with wild-type EGFR tested by cobas test. FoundationOne CDx was used for comprehensive genomic profiles. We then evaluated the missed EGFR mutations by the cobas test. We studied 62 patients. The median age was 60 (range: 35-86 years). Most patients were male and 58.1% were smokers. 91.9% were adenocarcinomas. Of the 62 samples, 7 (11.3%) were detected with EGFR mutations by NGS. Among these overlooked EGFR mutations, five were exon 20 insertions, and two were exon 19 deletions. Two patients received EGFR TKIs and showed durable response with PFS 5.9 months and 10.1 months, respectively. Using NGS as the standard, the false-negative rate of the cobas EGFR mutation test was 11.3%-in a population with a high prevalence of EGFR mutations. The most overlooked mutations were exon 20 insertions. A comprehensive EGFR mutation assay can provide significant benefits to patients with NSCLC.

Artificial Intelligence for Early Detection of Chest Nodules in X-ray Images.

Early detection increases overall survival among patients with lung cancer. This study formulated a machine learning method that processes chest X-rays (CXRs) to detect lung cancer early. After we preprocessed our dataset using monochrome and brightness correction, we used different kinds of preprocessing methods to enhance image contrast and then used U-net to perform lung segmentation. We used 559 CXRs with a single lung nodule labeled by experts to train a You Only Look Once version 4 (YOLOv4) deep-learning architecture to detect lung nodules. In a testing dataset of 100 CXRs from patients at Taipei Veterans General Hospital and 154 CXRs from the Japanese Society of Radiological Technology dataset, the sensitivity of the AI model using a combination of different preprocessing methods performed the best at 79%, with 3.04 false positives per image. We then tested the AI by using 383 sets of CXRs obtained in the past 5 years prior to lung cancer diagnoses. The median time from detection to diagnosis for radiologists assisted with AI was 46 (3-523) days, longer than that for radiologists (8 (0-263) days). The AI model can assist radiologists in the early detection of lung nodules.