Machine learning prediction of pathological complete response and overall survival of breast cancer patients in an underserved inner-city population.

BACKGROUND: Generalizability of predictive models for pathological complete response (pCR) and overall survival (OS) in breast cancer patients requires diverse datasets. This study employed four machine learning models to predict pCR and OS up to 7.5 years using data from a diverse and underserved inner-city population. METHODS: Demographics, staging, tumor subtypes, income, insurance status, and data from radiology reports were obtained from 475 breast cancer patients on neoadjuvant chemotherapy in an inner-city health system (01/01/2012 to 12/31/2021). Logistic regression, Neural Network, Random Forest, and Gradient Boosted Regression models were used to predict outcomes (pCR and OS) with fivefold cross validation. RESULTS: pCR was not associated with age, race, ethnicity, tumor staging, Nottingham grade, income, and insurance status (p > 0.05). ER-/HER2+ showed the highest pCR rate, followed by triple negative, ER+/HER2+, and ER+/HER2- (all p < 0.05), tumor size (p < 0.003) and background parenchymal enhancement (BPE) (p < 0.01). Machine learning models ranked ER+/HER2-, ER-/HER2+, tumor size, and BPE as top predictors of pCR (AUC = 0.74-0.76). OS was associated with race, pCR status, tumor subtype, and insurance status (p < 0.05), but not ethnicity and incomes (p > 0.05). Machine learning models ranked tumor stage, pCR, nodal stage, and triple-negative subtype as top predictors of OS (AUC = 0.83-0.85). When grouping race and ethnicity by tumor subtypes, neither OS nor pCR were different due to race and ethnicity for each tumor subtype (p > 0.05). CONCLUSION: Tumor subtypes and imaging characteristics were top predictors of pCR in our inner-city population. Insurance status, race, tumor subtypes and pCR were associated with OS. Machine learning models accurately predicted pCR and OS.

Deep learning applications to breast cancer detection by magnetic resonance imaging: a literature review.

Deep learning analysis of radiological images has the potential to improve diagnostic accuracy of breast cancer, ultimately leading to better patient outcomes. This paper systematically reviewed the current literature on deep learning detection of breast cancer based on magnetic resonance imaging (MRI). The literature search was performed from 2015 to Dec 31, 2022, using Pubmed. Other database included Semantic Scholar, ACM Digital Library, Google search, Google Scholar, and pre-print depositories (such as Research Square). Articles that were not deep learning (such as texture analysis) were excluded. PRISMA guidelines for reporting were used. We analyzed different deep learning algorithms, methods of analysis, experimental design, MRI image types, types of ground truths, sample sizes, numbers of benign and malignant lesions, and performance in the literature. We discussed lessons learned, challenges to broad deployment in clinical practice and suggested future research directions.

Machine learning with textural analysis of longitudinal multiparametric MRI and molecular subtypes accurately predicts pathologic complete response in patients with invasive breast cancer.

PURPOSE: To predict pathological complete response (pCR) after neoadjuvant chemotherapy using extreme gradient boosting (XGBoost) with MRI and non-imaging data at multiple treatment timepoints. MATERIAL AND METHODS: This retrospective study included breast cancer patients (n = 117) who underwent neoadjuvant chemotherapy. Data types used included tumor ADC values, diffusion-weighted and dynamic-contrast-enhanced MRI at three treatment timepoints, and patient demographics and tumor data. GLCM textural analysis was performed on MRI data. An extreme gradient boosting machine learning algorithm was used to predict pCR. Prediction performance was evaluated using the area under the curve (AUC) of the receiver operating curve along with precision and recall. RESULTS: Prediction using texture features of DWI and DCE images at multiple treatment time points (AUC = 0.871; 95% CI: (0.768, 0.974; p<0.001) and (AUC = 0.903 95% CI: 0.854, 0.952; p<0.001) respectively), outperformed that using mean tumor ADC (AUC = 0.850 (95% CI: 0.764, 0.936; p<0.001)). The AUC using all MRI data was 0.933 (95% CI: 0.836, 1.03; p<0.001). The AUC using non-MRI data was 0.919 (95% CI: 0.848, 0.99; p<0.001). The highest AUC of 0.951 (95% CI: 0.909, 0.993; p<0.001) was achieved with all MRI and all non-MRI data at all time points as inputs. CONCLUSION: Using XGBoost on extracted GLCM features and non-imaging data accurately predicts pCR. This early prediction of response can minimize exposure to toxic chemotherapy, allowing regimen modification mid-treatment and ultimately achieving better outcomes.

Internal mammary lymph nodal response to neoadjuvant chemotherapy on imaging and breast cancer prognosis.

Background: Pathologic complete response (pCR) to neoadjuvant chemotherapy (NAC) is an important prognostic indicator in breast cancer. Internal mammary lymph node involvement is not currently included in pCR determination, as sampling at the time of surgery is not routinely performed. Methods: Pre and post neoadjuvant chemotherapy MRI or PET/CT imaging response of the internal mammary lymph node chain was utilized as a surrogate to pCR and imaging data was correlated with patient outcomes. Results: Internal mammary lymph node response to NAC was associated with disease free survival over the course of this study, regardless of whether axillary nodal pCR was achieved. Conclusion: Internal mammary lymph nodal response to NAC is an important prognostic indicator. Potential use of internal mammary lymph node resolution as an imaging data input for AI models that predict pCR post-NAC may improve accuracy and other metrics in pCR prediction.

Mammographic Findings of Diffuse Axillary Tail Trabecular Thickening After the Second Booster of COVID-19 Vaccination.

Axillary adenopathy post-coronavirus disease 2019 (COVID-19) vaccination has been well-documented and is seen with other types of vaccinations. Isolated trabecular thickening on mammography, however, is singular to COVID-19 vaccination, which implies that this finding may result from a distinct pathophysiologic mechanism. Herein, we describe the first case of axillary tail trabecular thickening resulting from the second booster of the COVID-19 vaccination series. Both breast cancer and mastitis may present similar findings. Ipsilateral injection of COVID-19 vaccine/booster and spontaneous resolution on follow-up provide clues to the etiology. It has been hypothesized that proinflammatory conditions may predispose to axillary tail trabecular thickening on mammography post-COVID-19 vaccination. Proinflammatory conditions such as hypertension, obesity, and diabetes may also predispose to breast cancer, making this scenario even more of a diagnostic dilemma. This scenario would more likely be seen in lower socioeconomic communities, African Americans, and Hispanics, who demonstrate a higher prevalence of these diseases, and who are also more vulnerable due to health care disparities negatively affecting these groups. We discuss our case and the importance of this public health issue. Sequela of COVID vaccination and boosters will be encountered in the foreseeable future and could pose a diagnostic dilemma, thus potentially straining the healthcare system with unnecessary biopsies and patient anxiety if not recognized and appropriately managed.

Deep Learning Prediction of Pathologic Complete Response in Breast Cancer Using MRI and Other Clinical Data: A Systematic Review.

Breast cancer patients who have pathological complete response (pCR) to neoadjuvant chemotherapy (NAC) are more likely to have better clinical outcomes. The ability to predict which patient will respond to NAC early in the treatment course is important because it could help to minimize unnecessary toxic NAC and to modify regimens mid-treatment to achieve better efficacy. Machine learning (ML) is increasingly being used in radiology and medicine because it can identify relationships amongst complex data elements to inform outcomes without the need to specify such relationships a priori. One of the most popular deep learning methods that applies to medical images is the Convolutional Neural Networks (CNN). In contrast to supervised ML, deep learning CNN can operate on the whole images without requiring radiologists to manually contour the tumor on images. Although there have been many review papers on supervised ML prediction of pCR, review papers on deep learning prediction of pCR are sparse. Deep learning CNN could also incorporate multiple image types, clinical data such as demographics and molecular subtypes, as well as data from multiple treatment time points to predict pCR. The goal of this study is to perform a systematic review of deep learning methods that use whole-breast MRI images without annotation or tumor segmentation to predict pCR in breast cancer.

Invasive Lobular Cancer Arising in a Surgical Scar From Lumpectomy for a Previous Invasive Ductal Cancer of the Breast.

We describe a case of pathology-proven invasive lobular breast cancer (ILC) arising in a scar over 15 years after lumpectomy for previous invasive ductal carcinoma (IDC). The tumor was detected on screening mammography as a new focal asymmetry at the scar site and confirmed at diagnostic mammography. Ultrasound demonstrated an irregular, shadowing, hypoechoic mass at the scar site. Ultrasound-guided biopsy revealed poorly differentiated invasive lobular carcinoma. MRI and CT showed an irregular mass with pectoralis muscle invasion. Multimodality imaging findings are described. This is the first case to our knowledge reporting multimodality imaging findings of a breast cancer developing at the site of a surgical scar that is histologically different from the originally resected cancer.

Mammographic findings of diffuse axillary tail trabecular thickening following immunization with mRNA COVID-19 vaccines: Case series study.

Axillary lymphadenopathy has been reported after ipsilateral COVID-19 vaccination and can cause confusion for possible malignancy [1]. Intrinsic findings isolated to the breast has not been previously reported. This is the first case series of ipsilateral reversible changes of diffuse axillary tail trabecular thickening on screening mammography in totally asymptomatic patients in connection with COVID vaccination, 3 of which were isolated findings, confirmed by complete resolution of all imaging findings on follow up. In all instances, imaging was performed within 1 week of the first or third dose of an mRNA COVID-19 vaccine. These findings can be confused with breast cancer. Spontaneous resolution distinguishes vaccine-related findings from breast cancer.

Soft matrices downregulate FAK activity to promote growth of tumor-repopulating cells.

Tumor-repopulating cells (TRCs) are a tumorigenic sub-population of cancer cells that drives tumorigenesis. We have recently reported that soft fibrin matrices maintain TRC growth by promoting histone 3 lysine 9 (H3K9) demethylation and Sox2 expression and that Cdc42 expression influences H3K9 methylation. However, the underlying mechanisms of how soft matrices induce H3K9 demethylation remain elusive. Here we find that TRCs exhibit lower focal adhesion kinase (FAK) and H3K9 methylation levels in soft fibrin matrices than control melanoma cells on 2D rigid substrates. Silencing FAK in control melanoma cells decreases H3K9 methylation, whereas overexpressing FAK in tumor-repopulating cells enhances H3K9 methylation. Overexpressing Cdc42 or RhoA in the presence of FAK knockdown restores H3K9 methylation levels. Importantly, silencing FAK, Cdc42, or RhoA promotes Sox2 expression and proliferation of control melanoma cells in stiff fibrin matrices, whereas overexpressing each gene suppresses Sox2 expression and reduces growth of TRCs in soft but not in stiff fibrin matrices. Our findings suggest that low FAK mediated by soft fibrin matrices downregulates H3K9 methylation through reduction of Cdc42 and RhoA and promotes TRC growth.

Comparison of Insall-Salvati ratios in children with an acute anterior cruciate ligament tear and a matched control population.

OBJECTIVE: An anterior cruciate ligament (ACL) injury is an increasingly recognized cause of knee pain in young patients and in athletes in particular and can be detected non-invasively with MRI. Anecdotal information suggests that patella alta may be more common in the setting of this injury, but no study has documented this phenomenon. This study sought to test whether an ACL tear is associated with an increased Insall-Salvati ratio suggestive of patella alta. MATERIALS AND METHODS: Measurements of patellar height, patellar tendon length, and the Insall-Salvati ratio obtained from MRI of the knee were compared for 34 children with an arthroscopically proven ACL tear (mean age ± SD, 12.4 ± 1.4 years) and 36 control subjects with normal knee examinations (12.8 ± 2.1 years); these measurements were performed independently by two observers. RESULTS: Patellar tendon length (47.6 ± 6.6 mm vs 40.4 ± 5.7 mm) and patellar position calculated as the Insall-Salvati ratio (1.16 ± 0.16 vs 0.99 ± 0.14) were significantly greater in the knees with an ACL injury than in those without an internal injury, respectively, on MRI (p < 0.001). Patella length was not significantly different between the two groups (patients vs control subjects, 41.1 ± 4.2 mm vs 40.6 ± 2.7 mm, respectively; p = 0.523). There was good to excellent interobserver and intraobserver correlation for all measurements. CONCLUSION: There is a significant association between an ACL tear and increased patellar tendon length with a greater Insall-Salvati ratio. The mechanism for this finding is unclear, but this association provides support to suggest relative patella alta may be a risk factor for ACL injuries in pediatric patients.

Evaluation of Contrast Extravasation as a Diagnostic Criterion in the Evaluation of Arthroscopically Proven HAGL/pHAGL Lesions.

Purpose. The validity of preoperative MRI in diagnosing HAGL lesions is debated. Various investigations have produced mixed results with regard to the utility of MRI. The purpose of this investigation is to apply a novel method of diagnosing HAGL/pHAGL lesions by looking at contrast extravasation and to evaluate the reliability of such extravasation of contrast into an extra-articular space as a sign of HAGL/pHAGL lesion. Methods. We utilized specific criteria to define contrast extravasation. We evaluated these criteria in 12 patients with arthroscopically proven HAGL/pHAGL lesion. We also evaluated these criteria in a control group. Results. Contrast extravasation occurred in over 83% of arthroscopically positive cases. Contrast extravasation as a diagnostic criterion in the evaluation of HAGL/pHAGL lesions demonstrated a high interobserver degree of agreement. Conclusions. In conclusion, extra-articular contrast extravasation may serve as a valid and reliable sign of HAGL and pHAGL lesions, provided stringent criteria are maintained to assure that the contrast lies in an extra-articular location. In cases where extravasation is not present, the "J" sign, though nonspecific, may be the only evidence of subtle HAGL and pHAGL lesions. Level of Evidence. Level IV, Retrospective Case-Control series.

Lymphatic filariasis disseminating to the upper extremity.

Lymphatic filariasis is the most common cause of acquired lymphedema worldwide (Szuba and Rockson, 1998). It is endemic to tropical and subtropical regions, and its effects are devastating. With over 100 million infected persons, it ranks second only to leprosy as the leading cause of permanent and long-term disability. Wuchereria bancrofti is the etiologic agent in 90% of cases. There is a dearth of published MRI findings with pathologically proven active infections, making this entity even more of a diagnostic dilemma. Imaging may provide the first clue that one is dealing with a parasite and may facilitate proper treatment and containment of this disease. This is the first report of pathologic correlation with MRI findings in the extremity in active filariasis. The magnetic resonance images demonstrate an enhancing, infiltrative, mass-like appearance with partial encasement of vasculature that has not been previously described in filariasis. Low signal strands in T2-hyperintense dilated lymphatic channels are seen and may depict live adult worms. We hypothesize that the low signal strands correspond to the collagen rich acellular cuticle. This, in combination with the surrounding hyperintense T2 signal, corresponding to a dilated lymphatic channel, may provide more specific MRI findings for active nematodal infection, which can prompt early biopsy, pathological correlation, and diagnosis.

Prevalence of idiopathic cuneate gyrus herniation based on emergency room CT examinations.

Idiopathic brain herniation caused by a focal dural defect can be confused for a mass or post-traumatic herniation. The prevalence of idiopathic brain herniation has never been described. We sought to ascertain the prevalence of idiopathic cuneate gyrus herniation in a general emergency room (ER) population on computed tomography (CT) imaging. The purpose of this study is to elucidate cuneate gyrus herniation and differentiate it from other pathologic conditions such as mass or traumatic herniation and to provide its anatomical prevalence in an ER population. Consecutive emergency room CT scans of the brain were evaluated prospectively for cuneate gyrus herniation over a 1 year period by a neuroradiologist. Of 1,500 brain CT scans evaluated, 11 patients demonstrated idiopathic cuneus gyrus herniation. The prevalence was 0.73 %. CT manifestations are normal brain tissue herniating into the superior cerebellar cistern. Idiopathic brain herniation can be mistaken for a pathologic process. We found the prevalence of one such idiopathic brain herniation, involving the cuneus gyrus, to be exceedingly rare. CT imaging demonstrates normal brain tissue herniating beyond the dural boundary.

Nature abhors a vacuum: bilateral prominent rotator cable in bilateral congenital absence of the long head of the biceps tendon.

Bilateral agenesis of the long head of the biceps tendon is an exceedingly rare anomaly and can be a challenging diagnostic dilemma whose differential diagnosis includes tear. We present the third case of bilateral agenesis of the long head of the biceps tendon. Absent or shallow intertubercular sulcus is a constant finding and serves to differentiate this entity from tear. Our case is unique in that there was a radial ray anomaly, where prior reports of bilaterality did not demonstrate associated congenital anomalies. We also describe the newly reported finding of thickening of the rotator cable without the presence of rotator cuff tear. Thickening of the rotator cable may be an associated finding with agenesis of the biceps tendon.

Passive Posterior Tibial Subluxation on Routine Knee MRI as a Secondary Sign of PCL Tear.

The posterior drawer test is an accurate clinical test to diagnose posterior cruciate ligament (PCL), indicating laxity of the PCL that allows posterior tibial translation. This study aimed to determine whether posterior tibial translation relative to the femur on routine MRI could serve as an additional sign of PCL tear. Routine knee MRI in eleven patients (7 males, 4 females) with arthroscopically confirmed isolated PCL tears were reviewed independently by two musculoskeletal radiologists. Measurements of tibial translation were made in the medial and lateral compartments of patients and controls (10 males, 12 females) without clinical or MRI evidence of ligament injury. Significant medial compartment posterior tibial translation was present in patients with PCL tear compared to controls (+2.93 mm versus +0.03 mm, P = 0.002) with excellent interobserver agreement (intraclass correlation coefficient (ICC) = 0.94). No significant difference in lateral compartment tibial translation was observed (+0.17 mm versus -0.57 mm, P = 0.366) despite excellent interobserver agreement (ICC = 0.96). Posterior tibial translation in the midmedial compartment may be a secondary sign of isolated PCL tear on routine knee MRI with passive extension without manipulation or weight bearing. Additional work in a larger cohort may better address the accuracy of this finding.

Methionine and choline regulate the metabolic phenotype of a ketogenic diet.

Low-carbohydrate ketogenic diets are commonly used as weight loss alternatives to low-fat diets, however the physiological and molecular adaptations to these diets are not completely understood. It is assumed that the metabolic phenotype of the ketogenic diet (KD) is caused by the absence of carbohydrate and high fat content, however in rodents the protein content of KD affects weight gain and ketosis. In this study we examined the role of methionine and choline in mediating the metabolic effects of KD. We have found that choline was more effective than methionine in decreasing the liver steatosis of KD-fed mice. On the other hand, methionine supplementation was more effective than choline in restoring weight gain and normalizing the expression of several fatty acid and inflammatory genes in the liver of KD-fed mice. Our results indicate that choline and methionine restriction rather than carbohydrate restriction underlies many of the metabolic effects of KD.

Cervical carotid artery dissection: current review of diagnosis and treatment.

Carotid artery dissection is a cause of stroke, especially in young and middle-aged patients. A dissection occurs when there is an intimal tear or rupture of the vasa vasorum, leading to an intramural hematoma, which is thought to result from trauma or can occur spontaneously, and is likely multifactorial, involving environmental and intrinsic factors. The clinical diagnosis of carotid artery dissection can be challenging, with common presentations including pain, partial Horner syndrome, cranial nerve palsies, or cerebral ischemia. With the use of noninvasive imaging, including magnetic resonance and computed tomography angiography, the diagnosis of carotid dissection has increased in frequency. Treatment options include thrombolysis, antiplatelet or anticoagulation therapy, endovascular or surgical interventions. The choice of appropriate therapy remains controversial as most carotid dissections heal on their own and there are no randomized trials to compare treatment options.

Static and rotational step-and-shoot IMRT treatment plans for the prostate: a risk comparison study.

PURPOSE: For certain clinical applications, rotational intensity modulated radiation therapy (R-IMRT) techniques such as volumetric modulated arc therapy (VMAT) are capable of improved target dose coverage and shorter delivery time when compared to static, step-and-shoot IMRT. The authors performed a risk comparison study for two standard step-and-shoot IMRT plans and two step-and-shoot Rotational IMRT plans that were designed specifically for a Siemens Primus linear accelerator. METHODS: CT images of a RANDO phantom were used to generate R-IMRT and static IMRT plans. One simple and one complex prostate cases were created to investigate these techniques. The R-IMRT plans consisted of 72 single-segmented 6 MV beams, equally spaced with beam angle separations of 5°. The static IMRT plans employed seven multisegmented 6 MV beams. Both types of plans were optimized in Pinnacle(3) with the direct machine parameter optimization (DMPO) algorithm using the same set of optimization objectives. The plans were delivered to a RANDO phantom and thermoluminescent diode (TLD) dose measurements were performed at various locations throughout the phantom. Risk coefficients and organ weightings as defined by International Commission on Radiological Protection (ICRP) Publication 103 were used to calculate the resulting effective doses to various organs at risk, as well as the overall risk estimate for both techniques. RESULTS: For the simple prostate case, the R-IMRT technique provided a higher dose to organs at risk within the CT volume and a lower overall peripheral dose to remaining organs. The R-IMRT plan had a risk estimate of 4.56% when compared to the IMRT risk of 4.78%. For the complex prostate case, there was no significant difference in the lifetime risks of the IMRT (5.73%) and R-IMRT (5.74%) plans. CONCLUSIONS: R-IMRT is an approximation to VMAT and it was found that there is no clinically significant difference between lifetime risk estimates between R-IMRT and standard seven-beam IMRT for the prostate.

Magnetic resonance imaging of idiopathic herniation of the cuneus gyrus.

A 65-year-old woman presenting with chronic headaches and without overt visual symptomatology was found to have herniation of the cuneus gyrus into the superior cerebellar cistern. Only one prior case of idiopathic brain herniation has been described, in which the parahippocampal gyrus herniated into the ambient cistern. In that case a biopsy was performed as the herniation was mistaken for a tumor. We describe the features of idiopathic brain herniation that would mitigate the necessity for undergoing brain biopsy.