Interaction between clinicians and artificial intelligence to detect fetal atrioventricular septal defects on ultrasound: how can we optimize collaborative performance?

OBJECTIVES: Artificial intelligence (AI) has shown promise in improving the performance of fetal ultrasound screening in detecting congenital heart disease (CHD). The effect of giving AI advice to human operators has not been studied in this context. Giving additional information about AI model workings, such as confidence scores for AI predictions, may be a way of further improving performance. Our aims were to investigate whether AI advice improved overall diagnostic accuracy (using a single CHD lesion as an exemplar), and to determine what, if any, additional information given to clinicians optimized the overall performance of the clinician-AI team. METHODS: An AI model was trained to classify a single fetal CHD lesion (atrioventricular septal defect (AVSD)), using a retrospective cohort of 121 130 cardiac four-chamber images extracted from 173 ultrasound scan videos (98 with normal hearts, 75 with AVSD); a ResNet50 model architecture was used. Temperature scaling of model prediction probability was performed on a validation set, and gradient-weighted class activation maps (grad-CAMs) produced. Ten clinicians (two consultant fetal cardiologists, three trainees in pediatric cardiology and five fetal cardiac sonographers) were recruited from a center of fetal cardiology to participate. Each participant was shown 2000 fetal four-chamber images in a random order (1000 normal and 1000 AVSD). The dataset comprised 500 images, each shown in four conditions: (1) image alone without AI output; (2) image with binary AI classification; (3) image with AI model confidence; and (4) image with grad-CAM image overlays. The clinicians were asked to classify each image as normal or AVSD. RESULTS: A total of 20 000 image classifications were recorded from 10 clinicians. The AI model alone achieved an accuracy of 0.798 (95% CI, 0.760-0.832), a sensitivity of 0.868 (95% CI, 0.834-0.902) and a specificity of 0.728 (95% CI, 0.702-0.754), and the clinicians without AI achieved an accuracy of 0.844 (95% CI, 0.834-0.854), a sensitivity of 0.827 (95% CI, 0.795-0.858) and a specificity of 0.861 (95% CI, 0.828-0.895). Showing a binary (normal or AVSD) AI model output resulted in significant improvement in accuracy to 0.865 (P < 0.001). This effect was seen in both experienced and less-experienced participants. Giving incorrect AI advice resulted in a significant deterioration in overall accuracy, from 0.761 to 0.693 (P < 0.001), which was driven by an increase in both Type-I and Type-II errors by the clinicians. This effect was worsened by showing model confidence (accuracy, 0.649; P < 0.001) or grad-CAM (accuracy, 0.644; P < 0.001). CONCLUSIONS: AI has the potential to improve performance when used in collaboration with clinicians, even if the model performance does not reach expert level. Giving additional information about model workings such as model confidence and class activation map image overlays did not improve overall performance, and actually worsened performance for images for which the AI model was incorrect. © 2024 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

Enteral feeding in duct-dependent congenital heart disease.

BACKGROUND: Infants with duct-dependent congenital heart lesions are treated with a prostaglandin E1 infusion. We aimed to describe the feeding strategies used at our institution in such infants, and to describe the incidence of necrotising enterocolitis (NEC) in this patient group, investigating whether enteral feeding is associated with a higher risk. METHODS: Patients diagnosed with hypoplastic left heart syndrome, coarctation of the aorta, pulmonary atresia, or transposition of the great arteries born over a defined period were identified. Premature infants, those with pre-existing gastrointestinal disease, and those who never received prostaglandin were excluded. Data were compared using univariable and multivariable logistic regression models. RESULTS: A total of 177 patients were identified, of them 18 received a diagnosis of suspected or confirmed NEC. There was no association between the diagnosis of NEC and enteral feeding (P = 0.9). CONCLUSIONS: Based on these data, there does not appear to be an association between enteral feeding and NEC in infants receiving prostaglandin.

Pubic osteomyelitis due to anaerobic bacteria.

Osteomyelitis of the pubic bone due to anaerobic bacteria has been reported infrequently, although an entity known as "sterile" osteitis pubis is common to the literature. We have described two cases of pubic osteomyelitis due to anaerobic bacteria, discussed two previously reported cases, and suggested that most cases of what has previously been termed sterile osteitis pubis may actually have been due to anaerobic bacteria that were not isolated because of deficiencies in collection, transport, and culture of clinical specimens. Included are the pathogenesis and an approach to the treatment of this entity.

Increased expression of luteinizing hormone/human chorionic gonadotropin receptor gene in human endometrial carcinomas.

Normal human endometrium expresses LH/hCG receptor gene. In the present study, we investigated whether human endometrial carcinomas also express this receptor gene. Reverse transcription-nested polymerase chain reaction amplified LH/hCG receptor sequences from human endometrial carcinoma just as it did those from normal human endometrium and human ovary as a positive control tissue. Northern blotting demonstrated that endometrial carcinomas contain a greater abundance of multiple LH/hCG receptor transcripts, which increased with increasing tumor grade. Western immunoblotting revealed that all grades of endometrial carcinomas contain multiple immunoreactive receptor proteins in greater abundance than normal endometrium. In situ hybridization and immunocytochemistry demonstrated not only the presence, but also higher LH/hCG receptor messenger ribonucleic acid and receptor protein levels in glands of endometrial carcinoma compared to glands in normal endometrium. Ligand blotting demonstrated that the 35-kilodalton protein receptor could bind [125I]hCG and that this binding was inhibited by excess unlabeled hCG. The binding was higher in endometrial carcinoma than in normal endometrium. Atrophic and endocervical glands from endometrial carcinoma samples contained very few or no receptors. In summary, our results demonstrate that human endometrial carcinomas not only contain but also appear to overexpress LH/hCG receptors compared to normal endometrium. This novel finding introduces previously unsuspected possibilities concerning the role of LH and its receptors in human endometrial carcinomas.

The expression of human chorionic gonadotropin/human luteinizing hormone receptors in human gestational trophoblastic neoplasms.

Normal human placental trophoblasts have recently been shown to contain receptors for hCG/hLH. The present studies investigated the expression of these receptors in hyperplastic and anaplastic trophoblasts in gestational trophoblastic neoplasms. The results demonstrated that both hydatidiform moles and choriocarcinomas contained receptor messenger RNA (mRNA) and receptor protein. A variety of nontrophoblast tumors, on the other hand, contained neither receptor mRNA nor receptor protein. Choriocarcinomas contained more receptor mRNA and receptor protein than hydatidiform moles which in turn contained more than normal human placenta. Midluteal phase human corpus luteum contained more receptor mRNA than normal human placenta and about the same as choriocarcinomas. The hyperplastic and anaplastic trophoblasts in hydatidiform moles and choriocarcinomas contained more receptor immunostaining than the normal trophoblasts in the same tissue or those from normal placentas from about the same gestational age. The receptor immunostaining increased as the degree of trophoblast hyperplasia increased in hydatidiform moles. Anaplastic trophoblasts of choriocarcinomas contained a similar amount of receptor immunostaining as severely hyperplastic trophoblasts of hydatidiform moles. Invading anaplastic trophoblasts of choriocarcinoma contained greater amount of receptor immunostaining than the surrounding endometrial stromal and myometrial smooth muscle cells. In summary, this is the first study to our knowledge demonstrating the expression of hCG/hLH receptor gene in gestational trophoblastic neoplasms. The increased receptor expression in these neoplasms suggests that hCG, via its receptors, could play a fundamental and previously unsuspected autocrine role in the regulation of trophoblast transformation, growth, invasion, and high hCG secretion.

Lack of self-regulation of human chorionic gonadotropin biosynthesis in human choriocarcinoma cells.

Human gestational trophoblastic neoplasms overexpress hCG/LH receptors. Whether this overexpression is a reflection of a loss of self-regulation of hCG biosynthesis was investigated using JAR human choriocarcinoma cells. The results show that exogenous hCG did not affect steady state hCG alpha and hCG beta mRNA or dimer hCG protein levels in JAR cells. The JAR cells, however, responded to 8-bromo-cAMP with an increase in hCG alpha mRNA levels, suggesting that cAMP-mediated regulation of the hCG subunit genes was intact in the cells. Disruption of receptor function by a receptor antibody, which resulted in an increase in hCG alpha mRNA levels and hCG secretion in normal trophoblasts, had no effect on JAR cells. Unlike normal trophoblasts, which contain a predominant receptor transcript of 1.8 kilobases (kb), with minor higher molecular size (7.5 and 5.4 kb) transcripts occasionally seen, JAR cells contain a higher abundance of multiple transcripts (7.5, 5.4, 3.5, and 1.8 kb), with the predominant transcript being 5.4 kb. In addition, although normal trophoblasts contain an 80-kilodalton receptor protein, JAR cells contain only a 50-kilodalton hCG/LH receptor isoform. In contrast to the effects of exogenous hCG on normal placental tissue in vitro, it was unable to down-regulate receptor transcripts or receptor protein in JAR cells. In summary, JAR cells lack the ability to self-regulate hCG biosynthesis. This loss could explain how hCG can reach very high levels in gestational trophoblastic disease compared to those in normal pregnancy.

Diagnosis and staging of ovarian carcinoma.

The early diagnosis of ovarian carcinoma will await the development of adequate and accurate screening tests. In the interval prior to development of these tests, ovarian cancers will frequently be discovered in an advanced stage. Improved methods of treatment must be based on an adequate trial of existing treatment methods founded on an understanding of factors influencing survival. FIGO stage grouping identifies many of these factors and should be used to identify patients with similar characteristics. Other prognostic factors such as tumor grade, volume of residual disease and the presence of ascites must be recorded and considered in evaluating therapeutic trials. The adoption of a routine method for patient evaluation and exploration may enhance the amount of information available for each patient and assure that adequate information is available with which to place the pateint in FIGO stage and substage.

Incidence of complications with mini vaginal culpostats in carcinoma of the uterine cervix.

Between 1980 and 1987, 298 patients with carcinoma of the uterine cervix were treated at the University of Louisville Department of Radiation Oncology. Of these, 197 (66.1%) were treated for cure by radiotherapy alone: 36 by external beam alone and 161 by external beam and tandem and ovoid applications. The F.I.G.O. staging of the 161 patients was 82 (50.1%) Stage IB, 9 (5.6%) Stage IIA, 40 (24.9%) Stage IIB, and 30 (18.6%) Stage III. The usual treatment was whole pelvis irradiation followed by two intracavitary applications using the Fletcher Suit Applicators of tandem and ovoids in 79/161 patients (49%), a 3-M Mini Applicator (Fletcher Suit Delcos Applicator) in 52/161 patients (32.3%), and a 3-M Mini Applicator with Caps in 30/161 patients (18.6%). The incidence of grade 3-4 gastrointestinal or genitourinary complications as defined by the RTOG was 19.3% (31/161). Various treatment parameters were analyzed to define possible contributing factors. Grade 3-4 complications were seen in 7.6% (6/79) of patients treated with the standard ovoid Fletcher system, 26.9% (14/52) treated with the mini-ovoid system, and 36.6% (11/30) treated with the mini-ovoid system with caps (p = .0006). Although trends were noted, neither the vaginal surface dose (VSD) from the ovoids nor the addition of the external beam dose to the VSD (total vaginal surface dose = TVSD) were significant independent variables (p = 0.19 and = 0.133, respectively). The TVSD was significant when comparisons were made between different ovoid systems (p = 0.05 for less than 12,000 cGy and p = 0.004 for greater than 12,000 cGy). In this study, the 3-M mini applicator was associated with a significant increase in grade 3-4 complications as compared to the Standard Fletcher Suit Applicator.

Presence and characteristics of receptors for [D-Trp6]luteinizing hormone releasing hormone and epidermal growth factor in human ovarian cancer.

This study was undertaken to establish the presence and characteristics of receptors for [D-Trp6]LH-RH on the membranes of human ovarian cancer. Specific binding of [125I, D-Trp6]LH-RH was found in 29 of 37 (78.4%) ovarian cancers and in 6 of 11 (54.5%) non-malignant human ovaries. Ligand binding was dependent on time and plasma membrane concentration in a fashion expected of a peptide hormone. Saturation, kinetic and displacement data were consistent with the presence of a highly specific, single class of non-cooperative binding site. On the basis of receptors affinity, LH-RH-receptor-positive ovarian cancers could be divided into two groups: high affinity group (Kd=2.71 +/- 0.60 nM; Bmax=0.46 +/- 0.07 pmol/mg membrane protein) comprising 55% of tumors, and low affinity group (Kd=78.0 +/- 19.6 nM; Bmax=9.44 +/- 2.68 pmol/mg membrane protein) which included 45% of tumors. LH-RH antagonist Cetrorelix showed an affinity to LH-RH receptors on ovarian cancers 14 times higher than the agonist [D-Trp6]LH-RH. Using 125I-epidermal growth factor, specific high affinity receptors were also detected in membranes from 13 of 24 (54%) ovarian cancers and 5 of 11 (45%) non-malignant ovaries. The demonstration of LH-RH receptors in human ovarian cancers provides a rationale for the use of therapeutic approaches based on LH-RH analogues in this malignancy. The probable involvement of growth factors in the development of ovarian cancers suggests the merit of trying a combined therapy based on analogs of LH-RH and somatostatin for this carcinoma.

Vulvovaginoplasty in gynecologic oncology.

Eight patients from M.D. Anderson Hospital underwent Williams' vulvovaginoplasty for sexual rehabilitation following pelvic exenteration. In 7 of the 8 patients the functional results were excellent, and none of the patients experienced morbidity. This surgical procedure is less extensive than other reconstructive methods and skin grafting or mechanical dilatation are rarely required.

Gonadal dysgenesis with vulvar carcinoma.

An unusual case of vulvar carcinoma in a gonadal dysgenesis patient is described. In addition to clinical manifestations, certain laboratory data related to endocrine status were gathered from serum and urine analyses. Finally, receptors for estrogens, progestins, and glucocorticoids were determined on biopsies of the vulvar carcinoma.

Preirradiation celiotomy and extended field irradiation for invasive carcinoma of the cervix.

Epidermoid carcinoma of the cervix was treated in 120 patients by means of exploratory celiotomy (with semitherapeutic excision of involved lymph nodes) followed by irradiation. The size of the field used for irradiation was determined by the presence and site of lymph involvement. Of 64 patients, metastatic cancer in pelvic nodes was found in 40, and in common iliac or aortic nodes in 24. Of the 2 groups, 8 and 3 patients, respectively, survived for 2 years or more. Irradiation to extended fields (using 5500 rads at 850 rads per week) controlled the cancer satisfactorily within the treated area, but the incidence of bowel complications was high. Recurrent carcinoma usually appeared as distant metastases outside the treatment area which suggests that patients with bulky primary lesions and positive nodes actually already have systemic disease as treatment is started. A safe yet effective dosage level for radiation therapy to extended fields has not yet been established.

Refinements in vaginal reconstruction using rectus abdominis flaps.

An improved method for vaginal reconstruction after pelvic exenteration or abdominal perineal resection is provided by the distally based rectus abdominis flap. This extended flap carries a skin paddle from the upper abdomen on the rectus abdominis muscle and deep inferior epigastric vascular pedicle. The skin paddle is inversely tubed to form a vaginal pouch and delivered transpelvically to the perineum. In addition to providing a vaginal reconstruction for sexual function, this reconstruction lessens pelvic wound complications in the exenteration patient by filling endopelvic dead space and revascularizing these frequently irradiated wounds. This method provides a neovagina with a single flap and does not involve additional donor sites in the thighs. Transpelvic passage from above not only fills endopelvic dead space better than thigh flaps, but also it allows retention of a vaginal cuff in supralevator resections. Another significant advantage of this reconstruction is its great reliability with minimal incidence of paddle loss. This flap design illustrates a type of flap refinement in which specific flaps can carry tissue from adjacent vascular territories because of anastomotic vessels between the two vascular territories, such as the vascular watershed between the deep inferior epigastric and superior epigastric vessels in this case.

Vaginal and pelvic reconstruction with distally based rectus abdominis myocutaneous flaps.

This report introduces a new method of vaginal reconstruction using a single rectus abdominis myocutaneous flap based distally. Applications of this flap in reconstruction of major abdominal wall and pelvic defects, such as hemipelvectomies, are also described. The flap is designed to carry a paddle of upper abdominal skin on a distally based muscle and vascular pedicle. Advantages of this flap design are (1) the technique is straightforward and rapid, (2) flap viability is reliable, (3) the epigastric skin-fascial donor defect preserves the anterior rectus fascia distal to the linea semicircularis, which prevents hernia, (4) a large arc of rotation is provided, and (5) the epigastric donor site does not interfere with colostomy and urinary conduit stomas in the pelvic exenteration patient. We have done 11 vaginal reconstructions and 9 major pelvic defect reconstructions with this flap during the last 3 1/2 years. In these 20 patients, the only complications were two partial flap losses. No major flap losses or ventral hernias occurred.

The use of laparoscopy to determine the results of chemotherapy for ovarian cancer.

The chemotherapeutic management of patients with epithelial tumors of the ovary is hampered by a lack of diagnostic tests that are sufficiently sensitive for detecting residual or recurrent tumor. At the completion of a chemotherapy program, direct visualization of the peritoneal cavity is necessary to accurately assess the amount and location of residual tumor. Laparoscopy, with visually directed biopsies of residual tumor masses, can spare some patients a laparotomy and yet provide a safe and easy route for assessment of intraabdominal disease. If residual tumor is not seen at laparoscopy, than laparotomy with multiple peritoneal biopsies must be done so that residual tumor will not be missed and chemotherapy prematurely stopped. Laparoscopy is not an alternative to "second look" laparotomy but is a useful adjunct for determining the presence of resectable, unresectable or diffuse disease after chemotherapy.

Retrospective analysis of patients with primary fallopian tube carcinoma treated at the University of Louisville.

Medical records of patients diagnosed with primary fallopian tube carcinoma between 1979 and 1989 were reviewed. Twenty-six patients were eligible; 8 patients were excluded after pathologic review, leaving 18 patients included in the study for this analysis. The median and mean age were 61 and 59 years, respectively, with a range of 39-80 years. There were three Stage I, five Stage II, seven Stage III, and three Stage IV patients. The most common presenting symptoms were abdominal/pelvic pain, abdominal distension, and vaginal discharge/bleeding. The primary site of the lesion was determined to be the right tube in 44% of the cases, the left tube in 39% of the patients, bilateral lesions in 11% of the patients, and indeterminate in 6%. Histologic grade was poorly differentiated (Grade III) in 13 patients, moderately differentiated (Grade II) in 4 patients, and well differentiated (Grade I) in one. No patient was correctly diagnosed preoperatively. Survival at 5 years of the entire group was 35% with a 3 year minimum followup. Corresponding disease free survival was 30%. Mean and median survival times were 74 and 37 months, respectively. The range of survival times was from 1 to 120 months. All Stage I patients, 80% (4/5) of Stage II, and 29% (2/7) of Stage III patients are alive without disease. None (0/3) of the Stage IV patients are alive. Treatment regimens consisted of intraperitoneal P-32, external beam radiotherapy, and/or chemotherapy. Radiotherapy was associated with a low incidence of treatment-related complications, the majority being gastrointestinal related. There was one chemotherapy-related death. These patients and their treatment outcomes add to the data base of numerous previous reports on fallopian tube carcinoma. Stage I and II patients fared excellently with primary surgical and adjuvant therapy. While the prognosis of Stage III and IV patients is much worse, significant levels of long term survival can be achieved with aggressive treatment.