The goal of this observational diagnostic study is to find how well two evaluation methods used during medical thoracoscopy can help identify whether pleural disease is malignant or benign. The study focuses on adults undergoing thoracoscopy for undiagnosed pleural effusion or suspected pleural disease.

The main questions this study aims to answer are:

How accurate is the thoracoscopist's visual assessment of the pleura in predicting malignancy?

How accurate is Rapid On-Site Evaluation (ROSE) of touch-imprint cytology compared with final laboratory histopathology?

Researchers will compare thoracoscopic visual impressions with ROSE results to determine which method provides more reliable real-time diagnostic information.

Participants will:

Undergo medical thoracoscopy as part of their clinical evaluation.

Have pleural biopsy samples assessed on-site using ROSE.

Have standard histopathology testing performed for final diagnosis.

This study may help improve decision-making during thoracoscopy by identifying whether combining visual assessment with ROSE leads to faster and more accurate diagnosis of pleural disease.

Pleural effusion is a common clinical problem, and distinguishing malignant from benign causes is crucial for guiding treatment. Medical thoracoscopy allows direct visualization of the pleural surfaces and enables targeted biopsies. Although experienced clinicians often rely on the gross appearance of pleural abnormalities, visual assessment alone can be limited by overlap between malignant and benign features. Rapid On-Site Evaluation (ROSE) of touch-imprint cytology has emerged as a technique that may provide immediate microscopic information during the procedure, potentially improving diagnostic accuracy and reducing the need for repeat procedures.

This observational study prospectively evaluated the diagnostic performance of thoracoscopic visual inspection compared with ROSE during medical thoracoscopy. Thirty-three adult patients undergoing thoracoscopy for undiagnosed pleural effusion or suspected pleural disease were included. During the procedure, the thoracoscopist recorded a visual impression of whether pleural findings appeared malignant or benign. Touch-imprint slides were prepared from pleural biopsy specimens and assessed immediately by a trained cytopathologist using ROSE. All biopsy samples were subsequently analyzed by formal histopathology, which served as the reference standard.

The study measured sensitivity, specificity, accuracy, predictive values, receiver operating characteristic (ROC) curves, and agreement using Cohen's kappa. Visual assessment showed high sensitivity but limited specificity. ROSE demonstrated high overall accuracy and strong agreement with final histopathology, suggesting it may substantially improve real-time diagnostic confidence. By integrating both visual inspection and ROSE, clinicians may achieve more reliable intraoperative decision-making and potentially streamline patient management during thoracoscopy.

Inclusion criteria included adult patients (>18 years) with undiagnosed pleural effusion undergoing medical thoracoscopy.

Exclusion Criteria:

Exclusion Criteria, while Medical thoracoscopy is generally safe; patients with the following conditions were excluded in accordance with reported contraindications (Jin et al., 2020):

1. Absolute contraindications:

   • Refusal to participate in the study.
   • Extensive pleural adhesions (e.g., pleural fibrosis, post-infection, or prior pleurodesis).
   • Inadequate pleural space for thoracoscopy.

2. Relative contraindications**

   • Coagulopathy (platelet count <60 × 10⁹/L or INR >1.2), unless corrected prior to the procedure. The use of aspirin and clopidogrel was not considered a contraindication.
   • Inability to tolerate the lateral decubitus position.
   • Unstable hemodynamic status.
   • Severe hypoxemia requiring high-flow oxygen therapy.

Adult patients (>18 years) with undiagnosed exudative pleural effusion undergoing medical thoracoscopy.

Rationale: This study involves diagnostic observations collected during routine clinical care. Data contain potentially identifiable patient information and are intended for internal research use only