Apparent diffusion coefficient histogram in the differentiation of benign and malignant testicular tumors

Abstract

Purpose: This retrospective study assessed the value of histogram parameters of the apparent diffusion coefficient (ADC) map (HA) in differentiating between benign and malignant testicular tumors. We compared the diagnostic performance of two different volume-of-interest (VOI) placement methods: VOI 1, the entire tumor; VOI 2, the tumor excluding its cystic, calcified, hemorrhagic, and necrotic portions. Materials and methods: We retrospectively evaluated 45 patients with testicular tumors examined with scrotal contrast-enhanced magnetic resonance imaging. These patients underwent surgery with the pathological result of seven benign and 39 malignant tumors. We calculated the HA parameters, including mean, median, maximum, minimum, kurtosis, skewness, entropy, standard deviation (SD), mean of positive pixels, and uniformity of positive pixels by the two different VOI segmentation methods. We compared these parameters using the chi-square test, Mann-Whitney U test, and area under the receiver operating characteristic curve (AUC) to determine their optimal cut-off, sensitivity (Se), and specificity (Sp). Result: This study included 45 patients with 46 testicular lesions (seven benign and 39 malignant tumors), one of which had bilateral testicular seminoma. With the VOI 1 method, benign lesions had significantly lower maximum ADC (p = 0.002), ADC skewness (p = 0.017), and ADC variance (p = 0.000) than malignant lesions. In contrast, their minimum ADC was significantly higher ADC (p = 0.000). With the VOI 2 method, the benign lesions had significantly higher ADC SD (p = 0.048) and maximum ADC (p = 0.015) than malignant lesions. In contrast, their minimum ADC was significantly lower (p = 0.000). With the VOI 1 method, maximum ADC, ADC variance, and ADC skewness performed well in differentiating benign and malignant testicular lesions with cut-offs (Se, Sp, AUC) of 1846.000 (74.4%, 100%, 0.883), 39198.387 (79.5%, 85.7%, 0.868), and 0.893 (48.7%, 100%, 0.758). Conclusion: The HA parameters showed value in differentiating benign and malignant testicular neoplasms. The entire tumor VOI placement method was preferable to the VOI placement method excluding cystic, calcified, hemorrhagic, and necrotic portions in measuring HA parameters. Using this VOI segmentation, maximum ADC performed best in discriminating benign and malignant testicular lesions, followed by ADC variance and skewness.

Keywords: apparent diffusion coefficient maps; benign testicular tumor; diffusion-weighted imaging; histogram analysis; malignant testicular tumor; testicular cancer.

Figure 1

The VOI placement method in malignant tumors. (A) A tumor in the right testicle showed uneven hypointensity on the ADC map. (B) The VOI 1 method (red) covered the entire tumor, including its cystic and necrotic portions. (C) The VOI 2 method excluded the tumor's cystic and necrotic portions (arrow).

Figure 2

The VOI placement method in benign tumors: (A) A tumor in the left testicle showed slight hypointensity on the ADC map. (B) The VOI 1 method (red) covered the entire tumor, including its cystic and necrotic portions. (C) The VOI 2 method excluded the tumor's cystic and necrotic portions (arrow).

Figure 3

Representative histograms for one case in the benign and malignant tumor groups. (A) A histogram of a 25-year-old man with the histopathological result of an epidermoid cyst. (B) A histogram of a 44-year-old man with seminoma.

Figure 4

The ROC curves for the HA parameters with the VOI 1 placement method.

Figure 5

The ROC curves for the HA parameters with the VOI 2 placement method.