Set-Up Errors, Organ Motion, Tumour Regression and its Implications on Internal Target Volume-Planning Target Volume During Cervical Cancer Radiotherapy: Results From a Prospective Study.

AIMS: Uterocervical motions and organ filling during cervical cancer conformal radiotherapy is complex. This prospective, observational study investigated set-up margins (clinical target vo, ume [CTV] to planning target volume [PTV]) for pelvic nodal CTV and internal margin (CTV to internal target volume [ITV]) expansions for uterocervical movements during cervical cancer radiotherapy. MATERIALS AND METHODS: During cervical cancer radiotherapy, a daily kilovoltage, cone-beam computed tomography (CBCT) scan was acquired. Bony anatomy-based rigid co-registration and matching to vessels/pelvic nodal region was carried out to document shifts, errors (systematic and random) and to calculate CTV to PTV margins. Subsequently, soft-tissue matching was carried out at the mid-cervical region and uterine fundus to record shifts, errors and to calculate CTV to ITV margins. RESULTS: In 67 patients, 1380 CBCT scans were analysed. The mean (±standard deviation) couch shifts for CTV pelvic nodal region in all directions were within 4.5-5.3 mm, systematic and random errors 3.0-3.6 mm and set-up margins of within 10 mm (except anterior margin 10.3 mm). For the mid-cervical region, mean shifts were 4.5-5.5 mm, systematic and random errors 2-4 mm amounting to <10 mm internal margins (CTV-ITV for cervix) and for uterine fundus mean (±standard deviation) shifts were larger in the superior direction (12.1 mm) but 4.0-7.5 mm in other directions, systematic and random errors 2-7 mm amounting to anisotropic margins in various directions (10 mm in anterior-posterior and lateral directions, 12-20 mm in superior-inferior directions) (CTV-ITV for uterine fundus). CONCLUSION: Our study suggests anisotropic CTV to ITV and CTV to PTV margins for cervical cancer radiotherapy.

A Bio-inspired Algorithm based Multi-class Classification Scheme for Microarray Gene Data.

Microarray gene data is widely known for its high dimensionality and volume. The utilization of microarray gene data is increasing now-a-days, owing to the advancement of medical science. Microarray gene data helps in diagnosing diseases quite accurately. However, processing microarray gene data is difficult and is usually not understandable. Taking this challenge into account, this work presents a user-friendly rule based classification model, which is easily understandable and does not demand users to have prior knowledge. The classification rules are formed with the help of cuckoo search optimization algorithm and the rules are pruned by the associative rule mining. Finally, the classification is performed with the help of the pruned rules. The performance of the proposed approach is satisfactory in terms of accuracy, sensitivity, specificity and time consumption.