Deep-learning algorithm to detect fibrosing interstitial lung disease on chest radiographs.

BACKGROUND: Antifibrotic therapies are available to treat chronic fibrosing interstitial lung diseases (CF-ILDs), including idiopathic pulmonary fibrosis. Early use of these treatments is recommended to slow deterioration of respiratory function and to prevent acute exacerbation. However, identifying patients in the early stages of CF-ILD using chest radiographs is challenging. In this study, we developed and tested a deep-learning algorithm to detect CF-ILD using chest radiograph images. METHOD: From the image archive of Sapporo Medical University Hospital, 653 chest radiographs from 263 patients with CF-ILDs and 506 from 506 patients without CF-ILD were identified; 921 were used for deep learning and 238 were used for algorithm testing. The algorithm was designed to output a numerical score ranging from 0 to 1, representing the probability of CF-ILD. Using the testing dataset, the algorithm's capability to identify CF-ILD was compared with that of doctors. A second dataset, in which CF-ILD was confirmed using computed tomography images, was used to further evaluate the algorithm's performance. RESULTS: The area under the receiver operating characteristic curve, which indicates the algorithm's detection capability, was 0.979. Using a score cut-off of 0.267, the sensitivity and specificity of detection were 0.896 and 1.000, respectively. These data showed that the algorithm's performance was noninferior to that of doctors, including pulmonologists and radiologists; performance was verified using the second dataset. CONCLUSIONS: We developed a deep-learning algorithm to detect CF-ILDs using chest radiograph images. The algorithm's detection capability was noninferior to that of doctors.

Artificial intelligence-based technology for semi-automated segmentation of rectal cancer using high-resolution MRI.

AIM: Although MRI has a substantial role in directing treatment decisions for locally advanced rectal cancer, precise interpretation of the findings is not necessarily available at every institution. In this study, we aimed to develop artificial intelligence-based software for the segmentation of rectal cancer that can be used for staging to optimize treatment strategy and for preoperative surgical simulation. METHOD: Images from a total of 201 patients who underwent preoperative MRI were analyzed for training data. The resected specimen was processed in a circular shape in 103 cases. Using these datasets, ground-truth labels were prepared by annotating MR images with ground-truth segmentation labels of tumor area based on pathologically confirmed lesions. In addition, the areas of rectum and mesorectum were also labeled. An automatic segmentation algorithm was developed using a U-net deep neural network. RESULTS: The developed algorithm could estimate the area of the tumor, rectum, and mesorectum. The Dice similarity coefficients between manual and automatic segmentation were 0.727, 0.930, and 0.917 for tumor, rectum, and mesorectum, respectively. The T2/T3 diagnostic sensitivity, specificity, and overall accuracy were 0.773, 0.768, and 0.771, respectively. CONCLUSION: This algorithm can provide objective analysis of MR images at any institution, and aid risk stratification in rectal cancer and the tailoring of individual treatments. Moreover, it can be used for surgical simulations.

Diagnosis of prosthetic joint infection at the hip using the standard uptake value of three-phase 99mTc-hydroxymethylene diphosphonate SPECT/CT.

PURPOSE: To retrospectively investigate whether the standard uptake value (SUV) of 99mTc-bone single-photon emission computed tomography (SPECT)/CT could be useful for predicting prosthetic joint infection (PJI) at the hip. PATIENTS AND METHODS: We analyzed the cases of 37 patients with a suspected PJI at the hip who underwent 99mTc-bone SPECT/CT and surgical intervention with pathological and bacterial examinations. We divided the cases into those with and those without a causative bacterium detected in a surgical specimen, i.e., the positive bacterial culture (PBC) group (n = 17) and negative bacterial culture (NBC) group (n = 20). Cases with neutrophilic infiltration of surgical specimen comprised the positive neutrophilic infiltration (PINF) group (n = 18) and those without INF comprised the non-neutrophilic infiltration (NINF) group (n = 19). Quantitative analyses were performed using maximum SUVs and peak SUVs of blood-pool (BP) phase images (SUVmaxBP and SUVpeakBP) and late (LT)-phase images (SUVmaxLT and SUVpeakLT). RESULTS: Regarding the bacterial cultures, there were significant differences between the PBC and NBC groups in SUVmaxBP (5.26 ± 1.49 vs. 4.21 ± 1.15, respectively; p = 0.019), SUVpeakBP (4.89 ± 1.32 vs. 3.87 ± 1.06, p = 0.012), SUVmaxLT (16.10 ± 6.36 vs. 11.67 ± 4.95, p = 0.026), and SUVpeakLT (14.58 ± 5.83 vs. 10.49 ± 4.31 p = 0.036). Regarding neutrophilic infiltration, there were significant differences between the PINF and NINF groups in SUVmaxBP (5.18 ± 1.48 vs. 4.24 ± 1.19, p = 0.047) and SUVpeakBP (4.78 ± 1.32 vs. 3.92 ± 1.10, p = 0.043). CONCLUSION: An SUV analysis of 99mTc-bone SPECT/CT is a useful method to differentiate a PJI at the hip from non-infection.

DWI-related texture analysis for prostate cancer: differences in correlation with histological aggressiveness and data repeatability between peripheral and transition zones.

BACKGROUND: We investigated the correlation between texture features extracted from apparent diffusion coefficient (ADC) maps or diffusion-weighted images (DWIs), and grade group (GG) in the prostate peripheral zone (PZ) and transition zone (TZ), and assessed reliability in repeated examinations. METHODS: Patients underwent 3-T pelvic magnetic resonance imaging (MRI) before radical prostatectomy with repeated DWI using b-values of 0, 100, 1,000, and 1,500 s/mm2. Region of interest (ROI) for cancer was assigned to the first and second DWI acquisition separately. Texture features of ROIs were extracted from comma-separated values (CSV) data of ADC maps generated from several sets of two b-value combinations and DWIs, and correlation with GG, discrimination ability between GG of 1-2 versus 3-5, and data repeatability were evaluated in PZ and TZ. RESULTS: Forty-four patients with 49 prostate cancers met the eligibility criteria. In PZ, ADC 10% and 25% based on ADC map of two b-value combinations of 100 and 1,500 s/mm2 and 10% based on ADC map with b-value of 0 and 1,500 s/mm2 showed significant correlation with GG, acceptable discrimination ability, and good repeatability. In TZ, higher-order texture feature of busyness extracted from ADC map of 100 and 1,500 s/mm2, and high gray-level run emphasis, short-run high gray-level emphasis, and high gray-level zone emphasis from DWI with b-value of 100 s/mm2 demonstrated significant correlation, excellent discrimination ability, but moderate repeatability. CONCLUSIONS: Some DWI-related features showed significant correlation with GG, acceptable to excellent discrimination ability, and moderate to good data repeatability in prostate cancer, and differed between PZ and TZ.

Abnormal [18F]fluorodeoxyglucose accumulation to tori tubarius in IgG4-related disease.

OBJECTIVES: Tubarial glands (TGs) are recently refocused gland tissues localized near the tori tubarius in the nasopharynx and their clinical relevance is not clear yet. IgG4-related disease (IgG4-RD) is a progressive fibrosing condition and salivary glands are well-affected lesions. The aim of the present study is to examine [18F]fluorodeoxyglucose ([18F]FDG) accumulation to the tori tubarius in IgG4-related disease (IgG4-RD). METHODS: 48 patients with IgG4-RD who underwent positron emission tomography (PET) scanning with [18F]FDG were included and semi-quantitative analysis of [18F]FDG accumulation to tori tubarius was performed along with the clinical features and histopathological analysis. RESULTS: Of the 48 patients, abnormal [18F]FDG accumulation (metabolic tumour volume ≥ 1) to tori tubarius was observed in 15 (31.3%), all of whom had lesions in other head and neck glands. IgG4-RD patients with abnormal [18F]FDG accumulation to tori tubarius showed swollen nasopharyngeal walls around tori tubarius and forceps biopsy of the lesion revealed acinar cells and IgG4-positive plasma cells histologically. Abnormal [18F]FDG accumulation (maximum standard uptake value, metabolic tumour volume and total lesion glycolysis) to tori tubarius correlated with higher IgG4 and lower IgA serum concentrations. CONCLUSIONS: Abnormal [18F]FDG accumulation to tori tubarius can be observed in patients with IgG4-RD and the abnormal [18F]FDG accumulation to tori tubarius can be a clue of TG involvement in IgG4-RD.

MRI-detected extramural vascular invasion potentiates the risk for pathological metastasis to the lateral lymph nodes in rectal cancer.

PURPOSE: Multidisciplinary treatment for locally advanced rectal cancer requires an accurate assessment of the risk of metastasis to the lateral lymph nodes (LNs). We herein aimed to stratify the risk of pathological metastasis to lateral LNs based on the preoperatively detected malignant features. METHODS: All patients with rectal cancer who underwent surgery from January 2016 to July 2020 were identified. We recorded the TNM factors; perirectal and lateral LN sizes; and MRI findings, including mesorectal fascia involvement, extramural vascular invasion (EMVI), tumor site, and tumor distance from the anal verge. RESULTS: 101 patients underwent rectal resection with lateral lymph node dissection, of whom 16 (15.8%) exhibited pathological metastases to the lateral LNs. Univariate analyses demonstrated that lateral LN metastasis was significantly correlated with mrEMVI positivity (p = 0.0023) and a baseline lateral LN short-axis length of ≥ 5 mm (p < 0.0001). These significant associations were confirmed by a multivariate analysis (p = 0.0254 and 0.0027, respectively). The lateral LN metastasis rate was as high as 44% in cases bearing both risk factors, compared to 0% in cases lacking both risk factors. CONCLUSION: The results elucidated in this study may contribute to risk stratification, which can be used when determining the indications for lateral lymph node dissection.

CT reconstruction with thick slices not only underestimates lymph node size but also reduces data reproducibility in colorectal cancer.

BACKGROUND: Reliable size measurement of lymph node (LN) metastases is important for the evaluation of cancer treatment. However, image analyses without proper settings may result in inappropriate diagnoses and staging. PURPOSE: To investigate whether reconstruction slice thickness in computed tomography (CT) affects measurements of LN size and reproducibility. MATERIAL AND METHODS: We analyzed 48 patients with histological diagnoses of sigmoid colon and rectal cancer who underwent contrast-enhanced CT colonography as part of a surgical treatment preparation. A board-certified radiologist selected 106 LNs whose short-axis diameter was ≥5 mm on 1-mm-thick images; the short-axis diameters were measured on 1- and 5-mm-thick images by the radiologist and residents and compared using Wilcoxon matched-pairs signed rank test. Data variation and reproducibility were evaluated using the F test and Bland-Altman analysis. P<0.05 was considered significant. RESULTS: Short-axis diameters measured on 5-mm-thick images were significantly lower than those measured on 1-mm-thick images (P<0.01), even in the LNs whose short-axis diameters were over twice the slice thickness (P<0.05). Of the 106 LNs, 57 showed short-axis diameter <5 mm on 5-mm-thick images; the maximum short-axis diameter was 6.7 mm on a 1-mm thick image. Data variation was significantly larger on 5-mm thick images than 1-mm-thick images in small LNs (P<0.05) and reproducibility on 5-mm-thick images was inferior to that on 1-mm-thick images. CONCLUSION: Thick reconstruction slices in CT can result in an underestimation of LN size and reduce data reproducibility. When measuring LN size, a thin reconstruction slice would be recommended based on targeted LN size.

Lateral Position With Gantry Tilt Further Improves Computed Tomography Image Quality Reconstructed Using Single-Energy Metal Artifact Reduction Algorithm in the Oral Cavity.

OBJECTIVE: To assess the limitations of single-energy metal artifact reduction algorithm in the oral cavity and evaluate the availability of a solution by setting the patient in a lateral position (LP) with the use of a gantry tilt (GT). METHODS: We analyzed 88 patients with dental metals retrospectively in study 1, and 74 patients prospectively in study 2. Patients were classified: metal I with dental metals in 1 region, metal II in 2 regions, and metal III in 3 regions. Patients underwent neck computed tomography examinations in a supine position (SP) in study 1, and 2 positions, an LP with a GT and an SP, in study 2. All images were reconstructed with this algorithm. Image quality was scored using a 4-point scale: 1 = severe artifact, 2 = moderate artifact, 3 = slight artifact, 4 = no artifact. The scores were compared between metal I, metal II, and metal III using the Mann-Whitney U test in study 1, and between an LP with a GT and an SP using the Wilcoxon signed ranks test in study 2. RESULTS: The scores outside the dental arch were significantly higher in metal I than in metal II and metal III (3.0 ± 0.6 vs 2.3 ± 0.5 vs 2.2 ± 0.4; P < 0.0001 for metal I vs metal II and for metal I vs metal III) and significantly higher in an LP with a GT than an SP (3.2 ± 0.4 vs 2.3 ± 0.4; P < 0.0001). CONCLUSIONS: Single-energy metal artifact reduction algorithm could reduce metal artifacts adequately in patients with dental metals in 1 region, but not in 2 or more regions. However, even for the latter, combination of this algorithm and an LP with a GT could further improve the image quality.

Repeatability analysis of ADC histogram metrics of the uterus.

BACKGROUND: Recently, histogram analysis based on voxel-wise apparent diffusion coefficient (ADC) value distribution has been increasingly performed. However, few studies have been reported regarding its repeatability. PURPOSE: To evaluate the repeatability of ADC histogram metrics of the uterus in clinical magnetic resonance imaging (MRI). MATERIAL AND METHODS: Thirty-three female patients who underwent pelvic MRI including diffusion-weighted imaging (DWI) were prospectively included after providing informed consent. Two sequential DWI acquisitions with identical parameters and position were obtained. Regions of interest (ROIs) for histologically confirmed uterine lesions (five cervical and three endometrial cancers, and one endometrial hyperplasia) and normal appearing tissues (21 endometrium and 33 myometrium) were assigned on the first DWI dataset and then pasted onto the second DWI dataset. ADC histogram metrics within the ROIs were calculated and repeatability was evaluated by calculating within-subject coefficient of variance (%) (wCV (%)) and Bland-Altman plot (%). RESULTS: ADC 10%, 25%, median, 75%, 90%, maximum, mean, and entropy showed high repeatability (wCV (%) < 7, 95% limit of agreement in Bland-Altman plot (%) < ±20), followed by ADC minimum (wCV (%) = 8.12, 95% limit of agreement in Bland-Altman plot (%) < ±30). However, ADC skewness and kurtosis showed very low repeatability in all evaluations. CONCLUSION: ADC histogram metrics like ADC 10%, 25%, median, 75%, 90%, maximum, mean, and entropy are robust biomarkers and could be applicable to clinical use. However, ADC skewness and kurtosis lack robustness. Radiologists should keep these characteristics and limitations in mind when interpreting quantitative DWI.

Clinical utility of 18 F-fluorodeoxyglucose/positron emission tomography in diagnosis of immunoglobulin G4-related sclerosing sialadenitis.

OBJECTIVES/HYPOTHESIS: The aim of this study was to evaluate the utility of 18 F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) for accurately diagnosing immunoglobulin G4-related sclerosing sialadenitis (IgG4-SS). STUDY DESIGN: Retrospective cohort study. METHODS: We reviewed the records of 64 patients with IgG4-SS (35 male and 29 female patients) and 10 patients with clinically suspected IgG4-SS. Pathological diagnoses of patients clinically suspected with IgG4-SS included four cases of malignant lymphoma, one case of multicentric Castleman disease, one case of Sjögren's syndrome, and four cases of sialadenitis. All patients underwent submandibular gland (SMG) biopsies and baseline FDG-PET/CT evaluation. Clinical, serological, pathological, and PET/CT findings were analyzed. We also investigated maximum standardized uptake values (SUVmax) in the salivary glands of 15 patients with malignant disease of the salivary glands during the same period. RESULTS: Increased FDG uptake in the SMG and parotid gland was found in 63 (98%) and 23 (35%) patients with IgG4-SS, respectively. FDG uptake of the bilateral SMG and unilateral SMG was recorded in 57 patients (89%) and six patients (9%), respectively. Mean SUVmax in patients with malignant disease of the salivary glands was significantly higher than that in patients with IgG4-SS (P = .035). We defined a positive test for IgG4-SS diagnosis as high SMG FDG uptake and serum IgG4 level ≥135 mg/dL, resulting in a sensitivity, specificity, and accuracy of 96.9%, 90.0%, and 86.4%, respectively. CONCLUSIONS: FDG-PET/CT findings in combination with serological and clinical findings may have the capacity to diagnose IgG4-SS and lead to less-invasive biopsy procedures. LEVEL OF EVIDENCE: 4. Laryngoscope, 128:1120-1125, 2018.

Three cases of implantation of a SureScan® system and MRI for investigating causes of pain.

We report three cases of implantation of the SureScan® system and magnetic resonance imaging (MRI) for investigating causes of pain. Although there were metal-induced artifacts on the MR images of 2 patients, the artifacts did not affect the images of structures that needed to be assessed to make the diagnosis. The SureScan® system enabled patients implanted with spinal cord stimulation devices to undergo MRI.

Influence of the Different Primary Cancers and Different Types of Bone Metastasis on the Lesion-based Artificial Neural Network Value Calculated by a Computer-aided Diagnostic System, BONENAVI, on Bone Scintigraphy Images.

OBJECTIVES: BONENAVI, a computer-aided diagnostic system, is used in bone scintigraphy. This system provides the artificial neural network (ANN) and bone scan index (BSI) values. ANN is associated with the possibility of bone metastasis, while BSI is related to the amount of bone metastasis. The degree of uptake on bone scintigraphy can be affected by the type of bone metastasis. Therefore, the ANN value provided by BONENAVI may be influenced by the characteristics of bone metastasis. In this study, we aimed to assess the relationship between ANN value and characteristics of bone metastasis. METHODS: We analyzed 50 patients (36 males,14 females; age range: 87-42 yrs median age:72.5 yrs) with prostate, breast, or lung cancer who had undergone bone scintigraphy and were diagnosed with bone metastasis (32 cases of prostate cancer, nine cases of breast cancer, and nine cases of lung cancer). Those who had received systematic therapy over the past years were excluded. Bone metastases were diagnosed clinically, and the type of bone metastasis (osteoblastic, mildly osteoblastic, osteolytic, and mixed components) was decided visually by the agreement of two radiologists. We compared the ANN values (case-based and lesion-based) among the three primary cancers and four types of bone metastasis. RESULTS: There was no significant difference in case-based ANN values among prostate, breast, and lung cancers. However, the lesion-based ANN values were the highest in cases with prostate cancer and the lowest in cases of lung cancer (median values: prostate cancer, 0.980; breast cancer 0.909; and lung cancer, 0.864). Mildly osteoblastic lesions showed significantly lower ANN values than the other three types of bone metastasis (median values: osteoblastic,; 0.939 mildly osteoblastic; 0.788, mixed type; 0.991, and osteolytic. 0.969) The possibility of a lesion-based ANN value below 0.5 was %10.9 for bone metastasis in prostate cancer, %12.9 for breast cancer, and %37.2 for lung cancer. The corresponding possibility were %14.7 for osteoblastic metastases, %23.9 for mildly osteoblastic metastases, %7.14 for mixed-type metastases, and %16.0 for osteolytic metastases. CONCLUSION: The lesion-based ANN values calculated by BONENAVI can be influenced by the type of primary cancer and bone metastasis.

Improvement in automated quantitation of myocardial perfusion abnormality by using iterative reconstruction image in combination with resolution recovery, attenuation and scatter corrections for the detection of coronary artery disease.

OBJECTIVE: An iterative reconstruction method in combination with resolution recovery, attenuation and scatter corrections (IR-RASC) can improve image quality. It, however, is undetermined whether this technique can improve the detection of coronary artery disease (CAD) when automated quantitative analysis is used. This study evaluated diagnostic values of IR-RASC in combination with automated quantitative analysis in stress myocardial perfusion imaging (MPI) in the CAD detection. METHODS: This study enrolled consecutive 64 patients (mean age 66.2 ± 17.3 years, 39 males) who had undergone both 99mTc-labeled tetrofosmin stress MPI and coronary angiography within 3 months. Stress MPI abnormalities quantified as summed stress score (SSS), summed rest score (SRS) and summed difference score (SDS) by Heart Risk View-S (HRV-S) and Quantitative Perfusion SPECT (QPS) softwares using IR-RASC images were compared with those by using conventional filtered back-projection method (FBP) images and angiographic findings. RESULTS: Based on expert visual assessment, SSS and SRS by HRV-S/QPS softwares with IR-RASC were significantly lower than those by HRV-S/QPS softwares with FBP at mid- and basal left ventricular segments. Receiver-operating characteristics analysis showed that areas under the curve assessed by HRV-S (0.687) and QPS (0.678) with IR-RASC were nearly identical to those (0.717-0.724) by expert assessment with FBP, and were significantly (P < 0.05) greater than those by HRV-S (0.505) and QPS (0.522) with FBP. When HRV-S was used, the specificity and diagnostic accuracy of IR-RASC in the CAD detection were significantly greater than those of FBP: 90.3 versus 51.6%, P < 0.0001 and 79.7 versus 54.7%, P = 0.0027, respectively. Likewise, when QPS was used, the specificity and diagnostic accuracy of IR-RASC in the CAD detection were significantly greater than those of FBP: 80.6 versus 41.9%, P < 0.0001, and 78.1 versus 51.6%, P = 0.0018, respectively. There, however, were no significant differences in sensitivity between IR-RASC and FBP images. CONCLUSIONS: IR-RASC can improve diagnostic accuracy of the CAD detection using an automated scoring system compared to FBP, by reducing false positivity due to artefactual appearance.

Trials of vaccines for pancreatic ductal adenocarcinoma: Is there any hope of an improved prognosis?

Pancreatic tumors are chemoresistant and malignant, and there are very few therapeutic options for pancreatic cancer, as the disease is normally diagnosed at an advanced stage. Although attempts have been made to develop vaccine therapies for pancreatic cancer for a couple of decades, none of the resultant protocols or regimens have succeeded in improving the clinical outcomes of patients. We herein review vaccines tested within the past few years, including peptide, biological and multiple vaccines, and describe the three sets of criteria used to evaluate the therapeutic activity of vaccines in solid tumors.

The signal intensity ratio of the optic nerve to ipsilateral frontal white matter is of value in the diagnosis of acute optic neuritis.

OBJECTIVES: To evaluate the usefulness of the signal intensity ratio (SIR) of the optic nerve to the white matter (WM) on short tau inversion recovery (STIR) images to diagnose acute optic neuritis (AON). METHODS: The 405 consecutive patients with suspected orbital diseases underwent orbital magnetic resonance imaging (MRI) using a 3-T scanner between June 2008 and August 2011. Among them, 108 optic nerves (33 AON and 75 control) were retrospectively analysed. The averaged SIR (SIRave) and maximum SIR (SIRmax) were defined as the averaged signal intensity (SI) of the optic nerve divided by that of WM, and the maximum SI of the optic nerve divided by averaged SI of WM, respectively. These values were compared between AON and control using the Mann-Whitney U test. A P < 0.05 was considered statistically significant. RESULTS: SIRave and SIRmax were significantly (P < 0.001) higher in the AON compared to the control. At a cut-off SIRave value of 1.119, the sensitivity, specificity and accuracy were 0.939, 0.840, and 0.870; and at a cut-off SIRmax value of 1.281, these were 1.000, 0.720 and 0.806, respectively. CONCLUSION: The SIR of the optic nerve to WM on STIR images is of value in diagnosing AON. KEY POINTS: • We propose a method of diagnosing acute optic neuritis using 3-T MRI. • Our method is simple and objective and requires no novel imaging techniques. • Our method shows high diagnostic accuracy.

Risk of node metastasis of sentinel lymph nodes detected in level II/III of the axilla by single-photon emission computed tomography/computed tomography.

In breast cancer, single-photon emission computed tomography/computed tomography (SPECT/CT) shows the exact anatomical location of sentinel nodes (SN). SPECT/CT mainly exposes axilla and partly exposes atypical sites of extra-axillary lymphatic drainage. The mechanism of how the atypical hot nodes are involved in lymphatic metastasis was retrospectively investigated in the present study, particularly at the level II/III region. SPECT/CT was performed in 92 clinical stage 0-IIA breast cancer patients. Sentinel lymph nodes are depicted as hot nodes in SPECT/CT. Patients were divided into two groups: With or without hot node in level II/III on SPECT/CT. The existence of metastasis in level II/III was investigated and the risk factors were identified. A total of 12 patients were sentinel lymph node biopsy metastasis positive and axillary lymph node dissection (ALND) was performed. These patients were divided into two groups: With and without SN in level II/III, and nodes in level II/III were pathologically proven. In 11 of the 92 patients, hot nodes were detected in level II/III. There was a significant difference in node metastasis depending on whether there were hot nodes in level II/III (P=0.0319). Multivariate analysis indicated that the hot nodes in level II/III and lymphatic invasion were independent factors associated with node metastasis. There were 12 SN-positive patients followed by ALND. In four of the 12 patients, hot nodes were observed in level II/III. Two of the four patients with hot nodes depicted by SPECT/CT and metastatic nodes were pathologically evident in the same lesion. Therefore, the present study indicated that the hot node in level II/III as depicted by SPECT/CT may be a risk of SN metastasis, including deeper nodes.