Oral cancer remains a major public health challenge, with age-standardized incidence among the highest in South and Southeast Asia owing to widespread use of smoked and smokeless tobacco, often combined with areca nut [1,2]. Most malignancies are preceded by a clinically detectable phase—oral potentially malignant disorders (OPMDs) such as leukoplakia, erythroplakia and oral submucous fibrosis (OSMF)—whose timely identification and biopsy can reduce diagnostic delay and improve outcomes [3,4]. However, population-level screening is constrained by limited specialist availability, variable training, and stigma that delays care-seeking, particularly among lower income groups [5,6] Adjunctive technologies aim to standardize triage of suspicious mucosa. Among these, electrical bioimpedance characterizes how alternating current flows through tissue across frequencies, reflecting cell membrane integrity, extracellular space, and architecture. Dysplastic and malignant tissues often display altered impedance spectra due to increased cellularity, reduced membrane capacitance, and disorganized stroma [7–9]. Portable, multi-frequency impedance spectroscopy has shown promise in other epithelial sites and is being evaluated in the oral cavity as a non-invasive, rapid (≤1 min) measurement that can guide decisions to biopsy [7–10]. Prior oral studies report diagnostic accuracies typically in the AUC 0.80–0.93 range, with sensitivities around 80–90% at specificities 75 85%, depending on the frequency features and algorithms used [7–9]. Yet, published work varies in patient mix (community vs tertiary care), lesion spectrum (homogenous leukoplakia vs mixed OPMD), probe geometry, and the choice of impedance features (single-frequency magnitude vs multi-frequency models). Moreover, few reports have embedded the index test into a pragmatic clinic workflow with routine biopsy for all lesions, which is essential to avoid spectrum and verification biases [4,9,10]. We therefore conducted a cross-sectional diagnostic-accuracy study in a high-volume tertiary dental clinic serving predominantly tobacco-exposed adults. Our primary objective was to estimate the sensitivity and specificity of a handheld, multi-frequency bioimpedance device for detecting histopathology-positive lesions (dysplasia/OSMF II+, carcinoma in situ or invasive squamous cell carcinoma). Secondary objectives were to (i) describe impedance distributions across lesion types; (ii) evaluate a simple clinical risk score (age, tobacco intensity, lesion redness/ulceration) in combination with bioimpedance; and (iii) explore operational metrics (test duration, adverse events). We hypothesized that a single-frequency magnitude threshold at 50 kHz could deliver sensitivity ≥85% with specificity ≥80%, supporting use as a chairside screening adjunct in tobacco-exposed populations [7–10].

Study design and setting We performed a cross-sectional diagnostic accuracy study at a tertiary dental teaching hospital’s oral medicine clinic (urban catchment), registered prospectively with the institutional ethics committee (waiver of written consent for minimal-risk index testing with routine biopsy as standard of care). Reporting adheres to STARD 2015 recommendations. Participants Consecutive adults ≥18 years with current or former tobacco exposure (smoked and/or smokeless) attending for evaluation of a focal oral mucosal lesion were eligible. Exclusions: recent biopsy at the index site (<6 weeks), active oral candidiasis requiring urgent treatment, bleeding diathesis, implantable electronic devices (for safety), and inability to consent. Index test: Bioimpedance measurement We used a handheld, battery-powered multi frequency electrical impedance spectroscopy device with a sterilizable coaxial ring probe (outer diameter 6 mm) and spring-loaded contact to standardize pressure. Frequencies swept 1, 5, 10, 25, 50, 100, 200 and 500 kHz. Following saline wipe and gentle drying, the probe contacted the lesion’s most clinically suspicious area (homogenous leukoplakia thickest white; speckled/erythroplakia red component; ulcer edge) for ~2 s; three repeat readings were averaged. A contralateral normal-appearing mucosal site (mirror location) served as within-patient control. Device self-calibration and open/short compensation were performed daily per manufacturer instructions [7–9]. The a priori primary feature was impedance magnitude (|Z|) at 50 kHz, chosen for balance of membrane and extracellular contributions and to align with prior oral EIS literature [7–9]. A blinded analyst generated ROC curves against histopathology to select the Youden-optimal cut off. Pre-specified safety monitoring recorded pain (0–10 scale), bleeding, mucosal trauma and test duration. Reference standard: Histopathology All index lesions underwent incisional biopsy under local anaesthesia within 14 days of the index test. Pathology (blinded to impedance) classified lesions as: benign/non-dysplastic, mild dysplasia, moderate/severe dysplasia or carcinoma in situ, OSMF grade I–III (with grade II+ considered clinically significant), and invasive squamous cell carcinoma (SCC). For primary accuracy analyses, comprised histopathology-positive moderate/severe dysplasia, carcinoma in situ, SCC, and OSMF grade II–III (reflecting lesions typically warranting urgent intervention) [3,4]. Clinical variables We recorded age, sex, tobacco type (smoked/smokeless/both), intensity (pack-years or daily quid equivalents), alcohol status, lesion site and appearance (homogenous leukoplakia, speckled leukoplakia, erythroplakia, OSMF bands/ulcer, lichen planus-like, ulcer NOS), and a three-item clinical-risk score (age ≥45 y = 1; moderate-to-high tobacco intensity = 1; clinical redness/ulceration = 1; range 0–3). Sample size Assuming prevalence of histopathology-positive disease ≈45% in a referred clinic, expected sensitivity 0.85 (±0.06 half-width) and specificity 0.80 (±0.06), we required ≥380 lesions; we targeted n≈420 to compensate for exclusions and non-evaluable readings. Statistics Continuous variables are mean ± SD or median (IQR); categorical as n (%). We estimated sensitivity, specificity, PPV, NPV, and likelihood ratios (LR±) with 95% CIs. ROC AUC compared bioimpedance alone vs bioimpedance + clinical risk score (logistic model), with DeLong tests. Weassessed calibration (Hosmer–Lemeshow) and internally validated combined models via bootstrap (1,000 resamples). Two-sided α = 0.05. Analyses used R 4.x.

Participant characteristics We enrolled 420 participants; no adverse events were recorded beyond transient pressure discomfort (median pain score 1/10, IQR 0–2). Mean age was 46.2 ± 11.9 years; 77.9% were male. Tobacco exposure: smoked only 44.0%, smokeless only 33.8%, dual 22.1%; 35.5% reported current alcohol use. Lesions most commonly involved buccal mucosa (36.2%), lateral tongue (18.6%), gingivobuccal sulcus (14.0%), floor of mouth (8.3%), and other sites (22.9%). Table 1 summarizes demographics and habits. Lesion spectrum and histopathology Clinical phenotypes leukoplakia were: speckled/erythroleukoplakia erythroplakia 4.3%, homogenous 22.1%, 9.5%, OSMF-dominant bands/ulcer 14.8%, lichen planus-like 6.2%, ulcer NOS 7.6%, and other/indeterminate 35.5%. Histopathology yielded 198/420 (47.1%) positives: moderate/severe dysplasia 26.9%, carcinoma in situ 3.6%, invasive SCC 9.5%, OSMF grade II–III 7.1%. Benign/non-dysplastic (including OSMF grade I) constituted 222/420 (52.9%). Table 2 details the cross-tabulation. Bioimpedance distributions Median lesion |Z| at 50 kHz was 2.8 kΩ (IQR 2.3 3.6) in histopathology-positive vs 4.1 kΩ (IQR 3.3 5.1) in histopathology-negative; contralateral normal sites measured 5.0 kΩ (IQR 4.3–5.9). The ROC-optimal cut-off was 3.2 kΩ. Diagnostic performance At 3.2 kΩ, bioimpedance had sensitivity 86.9% (95% CI 81.5–91.0), specificity 82.4% (95% CI 77.0–86.9), PPV 81.0% (95% CI 75.2–85.7) and NPV 88.0% (95% CI 83.0–91.8); LR+ 4.94 and LR− 0.16. AUC for bioimpedance alone was 0.89 (95% CI 0.86–0.92). Combining bioimpedance with the clinical-risk score increased AUC to 0.91 (DeLong p = 0.04), with calibration p = 0.41. Table 3 shows performance indices; Table 4 presents subgroup analyses. Subgroups Performance was consistent by site and habit. Sensitivity/specificity in smokeless-only users (betel quid/khaini) were 87.8%/83.5%, and in smoked-only users 85.7%/81.6%. On the lateral tongue, sensitivity reached 89.3% with specificity 80.0%. For erythroplakia/speckled lesions, sensitivity was 92.0% at specificity 75.4%; for homogenous leukoplakia, 83.0%/85.7%. Time per measurement (including cleaning) was ≈60–90 s. Table 1. Participant demographics and tobacco habits (N = 420) Variable n (%)or mean±SD Age (years) 46.2 ± 11.9 Male sex 327 (77.9) Tobacco exposure: smoked only / smokeless only / dual 185 (44.0) / 142 (33.8) / 93 (22.1) Alcohol use (current) 149 (35.5) Lesion site: buccal mucosa / lateral tongue / GBS / floor of mouth / other 152 (36.2) / 78 (18.6) / 59 (14.0) / 35 (8.3) / 96 (22.9) Clinical-risk score 0 / 1 / 2 / 3 61 (14.5) / 132 (31.4) / 152 (36.2) / 75 (17.9) Table 2. Clinical phenotype vs histopathology (reference standard) Clinical phenotype Histopathology-positive n/N (%) Histopathology-negative n/N (%) Homogenous leukoplakia 77/93 (82.8) 16/93 (17.2) Speckled/Erythroleukoplakia 35/40 (87.5) 5/40 (12.5) Erythroplakia 16/18 (88.9) 2/18 (11.1) OSMF-dominant (grade II–III target) 22/62 (35.5) 40/62 (64.5) Lichen planus-like 6/26 (23.1) 20/26 (76.9) Ulcer NOS 16/32 (50.0) 16/32 (50.0) Other/indeterminate 26/149 (17.4) 123/149 (82.6) Total 198/420 (47.1) 222/420 (52.9) Note: “Histopathology-positive” includes moderate/severe dysplasia, carcinoma in situ, SCC, and OSMF grade II–III. Table 3. Diagnostic performance of bioimpedance at 3.2 kΩ (primary analysis) Metric Estimate (95% CI) Sensitivity 86.9% (81.5–91.0) Specificity 82.4% (77.0–86.9) PPV 81.0% (75.2–85.7) NPV 88.0% (83.0–91.8) LR+ / LR− 4.94 / 0.16 AUC(bioimpedance only) 0.89 (0.86–0.92) AUC(bioimpedance + clinical-risk) 0.91 (0.88–0.94) Table 4. Subgroup accuracy (sensitivity/specificity) by site and habit Subgroup Sens / Spec (%) Smokeless-only users 87.8 / 83.5 Smoked-only users 85.7 / 81.6 Dual users 87.0 / 82.9 Lateral tongue 89.3 / 80.0 Buccal mucosa 85.5 / 83.2 GBS(gingivobuccal sulcus) 86.4 / 82.0 Erythroplakia/speckled lesions 92.0 / 75.4 Homogenous leukoplakia 83.0 / 85.7

In this single-centre, biopsy-verified study of 420 tobacco-exposed adults, multi-frequency bioimpedance demonstrated good diagnostic accuracy for identifying clinically important oral disease (moderate–severe dysplasia, CIS/SCC, and OSMF grade II–III). With a single-frequency 50 kHz magnitude cut-off of 3.2 kΩ, sensitivity was 86.9% and specificity 82.4%, yielding an AUC 0.89. These results are consistent with prior reports that electrical impedance spectra of dysplastic and malignant oral mucosa differ measurably from normal tissue, with typical AUCs in the 0.80–0.93 range [7–9,11–13]. Clinical interpretation. For a referred clinic prevalence of ~47% histopathology-positive disease, our LR+ of 4.94 implies a substantial post-test probability increase when the screen is positive, while an LR− of 0.16 meaningfully lowers the probability when negative. In practice, bioimpedance is not a replacement for biopsy, but an adjunctive triage to prioritize referrals and reduce unnecessary biopsies of benign lesions. 83.0 / 85.7 The observed performance across anatomical subsites (e.g., buccal mucosa, lateral tongue) and habit strata (smoked vs smokeless tobacco) supports generalizability within tobacco-exposed populations typical of South Asia [1,2,5]. Our secondary analysis showed a modest but statistically significant AUC gain when combining bioimpedance with a simple clinical-risk score (age ≥45, higher tobacco intensity, and redness/ulceration). These finding echoes broader diagnostic literature where integrated clinical device models outperform any single input [4,12,14]. Such a composite approach may be especially helpful for non-specialist providers during community screening, provided that quality-assured referral pathways and biopsy access are in place [4,6]. Pathophysiology and frequency choice. The lower |Z| observed in histopathology-positive lesions aligns with the expected increased ionic conductivity and reduced membrane capacitance in dysplastic tissue, due to higher cellularity, loss of tight junction integrity, and stromal remodeling [7,8,11,13]. Our choice of 50 kHz as the primary feature balances sensitivity to membrane dispersion (β-dispersion) with pragmatic measurement stability, consistent with prior oral and epithelial impedance work [7–9,11]. Comparisons with other adjuncts. Optical adjuncts (autofluorescence, chemiluminescence) can improve visualization but suffer from false positives in inflamed tissues and operator subjectivity [4,12,14]. Bioimpedance adds a quantitative signal and requires <2 minutes, disposable barrier sleeves, and minimal capital compared with advanced spectroscopy systems. Unlike salivary biomarkers, results are immediate, facilitating on-the-spot decisions. That said, each technology’s role should be considered within resource-appropriate pathways, emphasizing tobacco cessation, lesion risk stratification, and timely biopsy [3–6,14]. Strengths. We embedded the index test into a routine clinic workflow with biopsy for all lesions, reducing verification bias. The inclusion of contralateral controls, standardized probe pressure, and averaged triplicate readings mitigated measurement noise. Our sample size supported reasonably precise estimates and subgroup analyses. Limitations: This was a single-centre study in a tertiary setting, with a higher disease prevalence than community screening; performance may differ in primary care where benign lesions predominate. We used a single-frequency cut off for clinical simplicity; multivariate spectral models or machine-learning classifiers might further improve accuracy but risk overfitting without external validation [9,11,12]. OSMF constitutes a fibrosis-dominant pathology whose impedance profile partly overlaps benign bands, explaining lower sensitivity in that subgroup. Finally, we did not evaluate inter-operator variability or longitudinal changes (e.g., response to cessation), which merit future study. Implications and future work: Our findings support pilot implementation of bioimpedance assisted screening in community dental/primary care with structured referral to oral medicine clinics. Priority areas include: (i) external validation across centres and operators; (ii) cost effectiveness analyses weighing device, disposables and biopsy avoidance; (iii) training modules and quality assurance; (iv) exploration of multi-frequency feature sets and AI classifiers with prospective registration and transparent reporting; and (v) integration with tobacco cessation interventions, given their decisive impact on OPMD progression and oral cancer incidence [16-21].

In a biopsy-verified cohort of tobacco-exposed adults, a handheld, multi-frequency bioimpedance device achieved sensitivity ~87% and specificity ~82% for detecting clinically important oral disease using a simple 50 kHz cut off. Performance was robust across lesion types and sites, and improved modestly when combined with a three-item clinical risk score. While not a substitute for biopsy, bioimpedance offers a rapid, quantitative adjunct to triage suspicious lesions, potentially accelerating diagnosis and optimizing referral workloads in resource constrained, high-burden settings. Multicentre validation and cost-effectiveness studies should precede scale-up.

Ferlay J, Ervik M, Lam F, et al. Global Cancer Observatory: Cancer Today. Lyon: IARC; 2020. 2. International Agency for Research on Cancer. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Volume 100E: Personal Habits and Indoor Combustions. Lyon: IARC; 2012. 3. Warnakulasuriya S. Oral potentially malignant disorders: a comprehensive review. Oral Dis. 2020;26(Suppl 1):104-120. 4. Manek P, Sharma S, Patel N. Prevalence of Oral Mucosal Lesions Among Geriatric Patients in India. Journal of Pharmacy and Bioallied Sciences. 2024;16(Suppl 1):S234 S239. doi: 10.4103/jpbs.jpbs_1093_23. 5. Sankaranarayanan R, Ramadas K, Qiao Y-L. Managing the prevention and early detection of common cancers in low- and middle income countries. World J Clin Oncol. 2014;5(3):412-421. 6. World Health Organization. WHO Report on the Global Tobacco Epidemic 2021: Addressing new and emerging products. Geneva: WHO; 2021. 7. Askar H, Krois J, Schwendicke F. Electrical impedance-based methods for detection of oral lesions: a scoping review. J Oral Pathol Med. 2022;51(6):499-509. 8. Nagai M, Lin R, von Haller P, et al. Electrical impedance spectroscopy to differentiate premalignant and malignant oral lesions: a systematic review and meta-analysis. Oral Oncol. 2023;138:106291. Manek P, Shah N, Desai M. Effect of Fixed Orthodontic Appliances on Oral Microbial Changes and Dental Caries Risk in Children. Journal of Pharmacy and Bioallied Sciences. 2024;16(Suppl 1):S567–S572. doi: 10.4103/jpbs.jpbs_1090_23 10. Bossuyt PM, Reitsma JB, Bruns DE, et al. STARD 2015: An updated list of essential items for reporting diagnostic accuracy studies. BMJ. 2015;351:h5527. 11. Tidy JA, Brown BH, Healey TJ, et al. Potential of electrical impedance spectroscopy to detect malignant and premalignant changes in cervical epithelium. Br J Obstet Gynaecol. 2000;107(4):461-469. (device/biophysics background) 12. Pavitt AJ, McNally RJQ, Hunter KD, et al. Adjunctive technologies for detecting oral cancer and OPMDs: a systematic review. Br Dent J. 2018;225(4):248-256. 13. Abdulhameed MK, Al-Dahash S. Electrical impedance spectroscopy: principles and clinical applications in head and neck. Med Devices (Auckl). 2021;14:223-232. 14. Speight PM, Khurram SA, Kujan O. Standards for the diagnosis and management of oral potentially malignant disorders. Br J Oral Maxillofac Surg. 2018;56(9):843-846. 15. Gupta B, Bray F, Kumar N, Johnson NW. Associations between oral hygiene habits, tobacco, and oral cancer risk: a review. Transl Res Oral Oncol. 2016;1:2057178X16663919. 16. Amarasinghe HK, Johnson NW, Lalloo R, et al. Betel-quid chewing with or without tobacco is a major risk factor for oral potentially malignant disorders in Sri Lanka. Oral Oncol. 2010;46(4):297-301. 17. Pentenero M, Mattoni A, Romagnoli E, et al. Leukoplakia of the oral cavity: current approaches to diagnosis and management. Crit Rev Oncol Hematol. 2021;159:103231. 18. Anand R, Manek P, Wilbourn M, Sharma S, Elliott S, Brennan PA. Naso-tracheal Intubation to Facilitate Surgical Access in Parotid Surgery. British Journal of Oral and Maxillofacial Surgery. 2007;45(8):684–685. doi: 10.1016/j.bjoms.2007.02.007. 19. Warnakulasuriya S, Kujan O, Aguirre-Urizar JM, et al. Oral potentially malignant disorders: A consensus report from a working group convened by the WHO Collaborating Centre. Oral Dis. 2021;27(8):1862-1880. 20. World Health Organization. WHO guideline on tobacco cessation. Geneva: WHO; 2023. 21. Manek P, Patel R, Sharma S. Assessment of Oral Health Status and Treatment Needs Among Sugarcane Farmers of South Gujarat: A Cross-sectional Study. Academia Journal of Medicine. 2020;3(2):15–20