Quality of voice after treatment for T1A glottic squamous cell carcinoma was studied in two matched groups of males treated either with CO2 laser cordectomi (n = 18, mean age 65.2 years) or with full dose radiotherapy (n = 18, mean age 65.1 years). All patients had histologically verified invasive squamous cell carcinoma. Fifteen male patients (mean age 63.9 years) without laryngeal disorders were used as controls. Voice recordings prior to treatment, and both at 3 months and at 2 years after completed treatment were analyzed. Acoustic measures of shimmer, jitter, breathiness, harmonic-to-noise ratio and fundamental frequency (F0) average were calculated with the Soundscope program. Two plain measures were also used: time required to read a running speech voice sample, and number of breaths. Perceptual voice analysis was performed blindly by two groups of listeners, Group A (4 experienced listeners) and Group B (4 naive listeners). Group A estimated quality of voice according to a modified GRBAS score, whereas Group B estimated Grade only. We found voice quality both at 3 months and at 2 years after radiotherapy to be significantly better than after laser treatment, as assessed by the acoustic variables breathiness, jitter, F0 average, running speech voice sample reading time and number of breaths. The perceptual variables Grade (Group A and B), Breathiness, Asthenia and Strain were also significantly better after radiotherapy.

Absolute voice rest is difficult to comply with. Neither short-term, nor long-term vocal recovery differed significantly between groups. Within-group comparisons showed significant improvements in vocal stamina, immediate recovery from vocal loading and self-assessments of voice problems only for the group with recommended relative voice rest.

No abstract

Multivariate analysis of fluorescence spectra could allow classification of laryngeal lesions in vivo with high sensitivity and specificity. PLS performs at least as well as PCA, and PLS-DA performs as well as logistic regression techniques on these data.