Yashas Ullas Lokesha

All Publications

• Deep learning for accurate tumour volume measurement and prediction of therapy response in paediatric osteosarcoma. European radiology von Krüchten, R., Barrow, M., Adams, L., Singh, S. B., Varniab, Z. S., Suryadevara, V., Ghimire, P., Pribnow, A., Qi, J., Applin, D., Lokesha, Y. U., Nernekli, K., Daldrup-Link, H. E. 2025

  Abstract

  To assess treatment response in osteosarcoma, two automated convolutional neural networks (CNNs) were developed to quantify tumour volumes and predict response to induction chemotherapy using histopathology as the reference standard.This retrospective, multicentre study included magnetic resonance imaging (MRI) scans from osteosarcoma patients acquired between January 2006 and July 2024. A 3D U-Net CNN segmented tumours and calculated volumes at baseline and post-chemotherapy. A second CNN predicted treatment response based on MRI-derived tumour volume changes using histopathologic necrosis (≥ 90%) as the reference standard. Both models were trained on 162 scans from 81 patients (Centre A) and validated on 40 scans from 20 patients (10 per centre) with Centre B as the external test set. Human readers measured 3D tumour diameters and volumes, compared with CNN-derived volumes using Spearman's correlation, Bland-Altman plots, and Dice coefficients. Prediction performance was assessed using accuracy, sensitivity, and specificity, with significance determined by agreement metrics.Patients from Centre A had a mean age of 15 ± 5 years (52 males), and from Centre B a mean age of 13 ± 0 years (8 males). CNN- and human-derived tumour volumes showed strong correlation (Centre A: r = 0.98, Centre B: r = 0.95; p < 0.001). Dice coefficients were 0.86 (Centre A) and 0.81 (Centre B), with median Hausdorff distances of 15.0 mm and 14.2 mm. The response prediction model classified 16/20 cases (80% accuracy) with 90% sensitivity and 70% specificity.CNN-derived tumour volume measurements were comparable to human assessments. CNN-based volume changes predicted histopathologic response to chemotherapy in paediatric osteosarcoma.Question Accurate, noninvasive assessment of treatment response in paediatric osteosarcoma is limited by its reliance on manual tumour measurements and post-surgical histopathology. Findings Automated deep learning accurately measured tumour volumes on MRI and predicted chemotherapy response with 80% accuracy, 90% sensitivity, and 70% specificity. Clinical relevance Automated deep learning enables accurate tumour volume assessment and prediction of chemotherapy response in paediatric osteosarcoma, offering a noninvasive tool to support and refine patient management.

  View details for DOI 10.1007/s00330-025-12115-w

  View details for PubMedID 41176552

  View details for PubMedCentralID 4486345

• Apparent diffusion coefficient can assist in differentiating between benign and malignant primary bone tumors in pediatric patients. Skeletal radiology Lokesha, Y. U., Singh, S. B., von Krüchten, R., Varniab, Z. S., Kumar, M., Suryadevara, V., Sarrami, A. H., Liang, T., Wong, J., Pribnow, A., Daldrup-Link, H. E. 2025

  Abstract

  To evaluate differences in apparent diffusion coefficient (ADC) values between benign and malignant primary pediatric bone tumors and to assess their diagnostic accuracy in differentiating these tumors.We retrospectively analyzed MRI scans of 96 pediatric patients (54 males, 42 females; mean age 12.97 ± 3.9 years) with primary bone tumors who underwent diffusion-weighted imaging, including 48 benign and 48 malignant tumors. We measured ADCmean, ADCmin, and ADCmax of the solid tumor part, carefully avoiding cystic, necrotic, or sclerosed tumor areas. The Wilcoxon rank-sum test was used to test the distributional difference of benign vs malignant tumors. ROC curve analysis was performed to assess the diagnostic accuracy. The optimal cutoff of ADC values to differentiate benign and malignant bone tumors was defined as the point at which the Youden index, the sum of sensitivity and specificity, was maximized.The median values of the ADCmean, ADCmin, and ADCmax for benign bone tumors [1.34 (1.13-1.83), 0.98 (0.73-1.34), and 1.80 (1.57-2.46) × 10-3mm2/s, respectively] were significantly higher compared to malignant bone tumors [0.93 (0.78-1.03), 0.59 (0.43-0.72), and 1.35 (1.22-1.66) × 10-3mm2/s, respectively; all p < 0.05]. ADCmean yielded the highest diagnostic accuracy, with an optimal cutoff of 1.04 (0.94-1.15) × 10-3mm2/s (sensitivity 77%, specificity 93%, AUC = 0.91). An ADCmin cutoff of 0.82 (0.65-0.98) × 10-3mm2/s resulted in a sensitivity of 87.5%, specificity of 70.0%, and AUC of 0.85. An ADCmax cutoff of 1.48 (1.18-1.78) × 10-3mm2/s achieved a sensitivity of 68%, specificity of 81%, and AUC of 0.80.ADCmean, ADCmin, and ADCmax differ significantly between benign and malignant pediatric bone tumors, and the ADCmean provides the highest diagnostic accuracy.

  View details for DOI 10.1007/s00256-025-05060-8

  View details for PubMedID 41160129

  View details for PubMedCentralID 11099578

• Utilization of Gallium-68 Fibroblast Activation Protein Inhibitor Positron Emission Tomography (⁶⁸Ga-FAPI PET) for Head and Neck Malignancies With Neck Imaging Reporting and Data System (Ni-RADS) Correlation. Cureus Bhat, R. R., Shivalingappa, S. S., Ashok, M., Kesari, A., Kedilaya, S., L, Y. U. 2025; 17 (2): e78476

  Abstract

  Background Head and neck cancers (HNCs) encompass a group of malignancies that arise in the mucosal surfaces of the oral cavity, pharynx, larynx, and other related structures. Advances in imaging modalities such as positron emission tomography-computed tomography (PET-CT) and magnetic resonance imaging (MRI) have improved tumor detection and staging, aiding in personalized treatment approaches. PET-CT is used to diagnose and stage various cancers. Interpretation of neck masses can be quite challenging, particularly in the context of prior surgery and radiotherapy. A standardized lexicon and risk classification system for interpreting images in patients treated for HNC has been developed by the American College of Radiology (ACR) Neck Imaging Reporting and Data System (Ni-RADS) Committee. While fluorine-18 fluorodeoxyglucose (18F-FDG) PET-CT is widely used to assess head and neck malignancies, we have chosen to employ gallium-68 fibroblast activation protein inhibitor (68Ga FAPI) PET-CT in the Ni-RADS category in this investigation because to its exclusive advantages over FDG. Methodology This was a non-funded retrospective-prospective study conducted in the Department of Radiodiagnosis at Healthcare Global Hospital, KR Road, Bangalore, following approval from the Institutional Ethical Committee. The period of observation for this study was January 2024 to June 2024. Patients with known HNC who were on follow-up and referred for a 68Ga FAPI PET-CT scan were included in the study. The Ni-RADS score was assigned, and histopathological correlation was performed. Descriptive statistics were used, and sensitivity and specificity were calculated. Results Out of the 41 cases selected for the study, all Ni-RADS 1 cases (100%, 6/6) were nonmalignant. Among Ni-RADS 2 cases 3 (37.5%) were malignant and 5 (62.5%) were nonmalignant. Nearly all Ni-RADS 3 cases (26/27, 96.3%) were malignant, indicating that higher Ni-RADS scores strongly correlate with malignancy. Recurrence is significantly associated with a Ni-RADS 3 score, whereas nonmalignancy is associated with a lower score (P < 0.001). Similar results were also seen in the case of nodal recurrences. This study showed that 68Ga FAPI PET-CT has a high sensitivity of 88.3% and specificity of 95.8% in identifying recurrent malignant and nonmalignant cases. Conclusions We can conclude that FAPI PET, which offers several advantages over FDG, can be effectively used in the Ni-RADS criteria for diagnosing HNC recurrences. The utilization of FAPI PET in conjunction with contrast-enhanced CT facilitates the identification of tumor morphological and metabolic features. However, further research and larger cohorts are needed to improve prediction accuracy and guide personalized treatment decisions.

  View details for DOI 10.7759/cureus.78476

  View details for PubMedID 40051952

  View details for PubMedCentralID PMC11883844

• Role of Magnetic Resonance Imaging in Morphometric Alterations of Corpus Callosum in Stroke Patients. Cureus Reddy, B. S., Naik, D., Sakalecha, A. K., L, Y. U., Uhasai, K., Mannan V, J. 2023; 15 (2): e35332

  Abstract

  Corpus callosum plays a role in interhemispheric integration, language, intelligence, and creativity of individuals, hence variations in corpus callosum size are seen in various neurological diseases such as Alzheimer's and bipolar affective disorder. While the dimensions differ based on gender, age, and ethnicity, pathological variations are seen with some diseases such as vascular dementia, leukoaraiosis, stroke, and carotid artery stenosis. This study was conducted to compare the morphometric alterations of the corpus callosum between normal subjects and stroke patients using magnetic resonance imaging (MRI).This was a case-control study conducted on 84 subjects divided into cases and control groups. The widths of the genu, body & splenium, and anterior-posterior (AP) diameter of the corpus callosum were measured and the values were compared among the two groups. Student's t-test and regression analysis were utilized for the analysis of data and p<0.05 was considered statistically significant.Sixteen patients (19.04%) belonged to the age range of 18-40 years, 32 (38.09%) belonged to the age range of 41-60 years and 36 (42.8%) belonged to the age group of >60 years. There was no discrepancy between cases and controls or between the age groups. The mean width of genu, body & splenium, and AP diameter was compared between normal individuals and stroke patients. It was noted to be significantly lesser in cases than in controls. The morphometric indices i.e., width of genu, body & splenium, and AP diameter of the corpus callosum in cases versus controls were noted to be 9.8 ± 1.2 vs. 10.27 ± 0.3 mm, p=0.12; 5.1±0.9 vs. 5.3±0.24 mm, p=0.25; 12.11 ± 9.65 vs. 12.52 ± 13.9 mm, p=0.04 (significant) and 71.22±3.1 vs. 72.32±1.2, p=0.23, respectively.This study showed that patients with stroke have a significant reduction in morphometric indices i.e., width of genu, body & splenium, and the AP diameter of the corpus callosum when compared to normal individuals.

  View details for DOI 10.7759/cureus.35332

  View details for PubMedID 36974258

  View details for PubMedCentralID PMC10038770

• Diagnostic utility of strain elastography in assessing median nerve changes among rheumatoid arthritis patients without symptoms of carpal tunnel syndrome: An analytical observational study INDIAN JOURNAL OF RHEUMATOLOGY Ullas, L., Rachegowda, N., Hariprasad, S. 2022; 17 (4): 334-339

  View details for DOI 10.4103/injr.injr_270_21

  View details for Web of Science ID 000922662900002

• The Evaluation of Variations in Patterns of Sphenoid Sinus Pneumatization Using Computed Tomography in a South Indian Population. Cureus Parameshwar Keerthi, B. H., Savagave, S. G., Sakalecha, A. K., Reddy, V., L, Y. U. 2022; 14 (3): e23174

  Abstract

  Background and objective Knowledge about sphenoid sinus pneumatization is critical for skull base surgeries and functional endoscopic sinus surgery (FESS) in order to avoid serious complications like postoperative meningitis, sinusitis, cerebrospinal fluid (CSF) rhinorrhea, and intracranial hematoma. In this study, we aimed to assess the proportion of anatomical variants in sphenoid sinus pneumatization and to determine the common sphenoid pneumatization pattern in a South Indian population. Methods This retrospective study was conducted over a period of six months from July 2019 to December 2019 among 573 patients who underwent non-contrast CT (NCCT) or contrast-enhanced CT (CECT) of the brain, paranasal sinuses (PNS), orbit, and face. Results Most of the patients were in the age group of 20-39 years. The male-to-female ratio was 2.45:1. Among the posterior extensions, the most common variant was type D, followed by type C, type B, and type A. Among the clival extensions, the most common variant was Cliv-A, followed by Cliv-B, Cliv-C, and Cliv-D. The most common lateral wall pneumatization was bilateral lateral wall pneumatization followed by unilateral sinus wall pneumatization. Lat-A was the most common lateral wall pneumatization pattern followed by Lat-D, Lat-B, and Lat-C. Conclusion Our study intends to classify the sphenoid sinus pneumatization pattern and identify the most common variant among them, thereby guiding the skull base and FESS surgeons in choosing the correct mode of the operative procedure and also anticipating and avoiding complications of surgery.

  View details for DOI 10.7759/cureus.23174

  View details for PubMedID 35433147

  View details for PubMedCentralID PMC9009218

• Severity of COVID-19 Infection Using Chest Computed Tomography Severity Score Index Among Vaccinated and Unvaccinated COVID-19-Positive Healthcare Workers: An Analytical Cross-Sectional Study. Cureus Ravindra Naik, B., Anil Kumar, S., Rachegowda, N., Yashas Ullas, L., Revanth, R. B., Venkata Sai Aluru, N. R. 2022; 14 (2): e22087

  Abstract

  Coronavirus disease 2019 (COVID-19) vaccines protect against severe illness. However, data on post-vaccination COVID-19 breakthrough infections are limited.An analytical cross-sectional study was conducted from May 2021 to July 2021 among 2043 COVID-19-positive healthcare workers who were divided into a vaccinated group (n=1010) and an unvaccinated group (n=1033). A pre-tested questionnaire was circulated among the healthcare workers using Google Forms. Chest computed tomography (CT) severity score was the primary outcome variable analyzed using coGuide.The average age of the study population was less than 45 years in both groups (43.05 ± 13.02 years). Most respondents (62%) were males. Hypertension (39%) and diabetes (33%) were the most common underlying diseases. Significant differences in age and cardiac disease were observed between the two groups (p = 0.07 and p <0.001, respectively). However, the difference was insignificant (p >0.05) for gender, hypertension, and diabetes. Most unvaccinated respondents had an increased CT severity score, and the difference between the studies groups was significant (p <0.001). Of the 1,010 vaccinated individuals, 382 (37.82%) received the only first vaccination dose, and 628 (62.18%) received both doses. The CT severity score decreased after receiving both vaccination doses, and the difference between CT severity score and vaccination status was significant (p <0.001).COVID-19 was mild in the vaccinated group. Chest CT severity score index can be considered an efficient tool in predicting prognosis and monitoring disease in patients with COVID-19 in India.

  View details for DOI 10.7759/cureus.22087

  View details for PubMedID 35295366

  View details for PubMedCentralID PMC8917791

• "SWASTHA-SHWASA": UTILITY OF DEEP LEARNING FOR DIAGNOSIS OF COMMON LUNG PATHOLOGIES FROM CHEST X-RAYS INTERNATIONAL JOURNAL OF EARLY CHILDHOOD SPECIAL EDUCATION Aishwarya, N., Veena, M. B., Ullas, Y. L., Rajasekaran, R. 2022; 14 (5): 1895-1905

  View details for DOI 10.9756/INTJECSE/V14I5.198

  View details for Web of Science ID 000834211600001