Computer-aided assessment of regional vascularity of thyroid nodules for prediction of malignancy

Abstract

Color Doppler vascular index (VI) was assessed alone and in combination with grey-scale ultrasound (GSU) in regionally subdivided thyroid nodules in diagnosing thyroid cancer. Color Doppler sonograms of 111 thyroid nodules were evaluated by a home-developed algorithm that performed “offsetting” (algorithm for changing the area of a region of interest, ROI, without distorting the ROI’s contour) and assessed peripheral, central and overall VI of thyroid nodules. Results showed that the optimum offset for dividing peripheral and central regions of nodule was 22%. At the optimum offset, the mean VI of peripheral, central, and overall regions of malignant nodules were significantly higher than those of benign nodules (26.5 ± 16.2%, 21.7 ± 19.6%, 23.8 ± 4.6% v/s 18.2 ± 16.7%, 11.9 ± 15.1% and 16.6 ± 1.8% respectively, P < 0.05). The optimum cut-off of peripheral, central, and overall VI was 19.7%, 9.1% and 20.2% respectively. When compared to GSU alone, combination of VI assessment with GSU evaluation of thyroid nodules increased the diagnostic accuracy from 58.6% to 79.3% (P < 0.05). In conclusion, a novel algorithm for regional subdivision and quantification of thyroid nodular VI in ultrasound images was established, and the optimum offset and cut-off were derived. Assessment of intranodular VI in conjunction with GSU can increase the accuracy in ultrasound diagnosis of thyroid cancer.

Diagnostic value of 3D power Doppler ultrasound in the characterization of thyroid nodules

“…However, considering the tumoral vascularity of thyroid and nonthyroid masses, we think that these results are inconsistent [13][14][15][16][24][25][26]. In a study using a computer-aided color Doppler software, Baig et al developed a new method that divided thyroid nodule into central and peripheral regions and computed Values are given as mean values ± SD; P: significance level for all pairs the regional and entire vascular index (VI) of the nodule [9]. They found that the mean VI of malignant nodules was significantly higher than that of benign nodules, in agreement with our study [9].…”

“…In a study using a computer-aided color Doppler software, Baig et al developed a new method that divided thyroid nodule into central and peripheral regions and computed Values are given as mean values ± SD; P: significance level for all pairs the regional and entire vascular index (VI) of the nodule [9]. They found that the mean VI of malignant nodules was significantly higher than that of benign nodules, in agreement with our study [9]. On the other hand, studies involving the characterization of thyroid nodules with CEUS in which the microvascular flow pattern is evaluated in tumoral lesions support our own findings [24][25][26].…”

“…Three-dimensional US, 3D PDUS, and computeraided color Doppler software methods provide a number of advantages over conventional techniques [8,9]. Three-dimensional US is more accurate than two-dimensional examination in the evaluation of anatomical structures and assessment of disease and provides data that can be subjected to repeated evaluation [8,9]. Three-dimensional power Doppler US is able to quantify vascularity within organs, tissues and, tumors [10,11].…”

Update on Ultrasound Diagnosis for Thyroid Cancer

“…34 Some studies have suggested that central vascularity is a feature for thyroid malignancy, whereas benign thyroid nodules tend to have peripheral vascularity ( Figure 12). 23,[35][36][37][38] However, other studies have noticed a higher association of central vascularity with benignity rather than malignant status of thyroid nodules. 31, 39,40 Shah et al 41 suggested that, when making a decision for thyroidectomy, intranodular vascularity on ultrasonography should not be used as an independent parameter to determine the risk of malignancy in patients with an atypical cytology result.…”

“…46 It has been reported that using objective and quantitative measurement of nodular vascularity would allow a more accurate prediction of thyroid malignancy. 47 In a recent study, Baig et al 37 proposed a novel method that used an automated computer-aided approach for regional segmentation of thyroid nodules. The vascularity indexes of the peripheral and central sections of the thyroid nodule were evaluated to differentiate benign and malignant lesions.…”