Dorella Franchi

Dorella Franchi

Auto-generated profile, based on publicly available information

Dorella Franchi

Preventive Gynecology Unit, Division of Gynecology, European Institute of Oncology IRCCS, Milan, Italy

Research Focus & Metrics

Academic Career Length

15Years (2012-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 July 2020 in BMJ
Verified
Journal Article
Open access
PDF file
BCBen Van Calster
Ben Van Calster
LVLil Valentin
Lil Valentin
WFWouter Froyman
Wouter Froyman
CLChiara Landolfo
Chiara Landolfo
JCJolien Ceusters
Jolien Ceusters
DTDirk Timmerman
Dirk Timmerman
Objective To evaluate the performance of diagnostic prediction models for ovarian malignancy in all patients with an ovarian mass managed surgically or conservatively. Design Multicentre cohort study. Setting 36 oncology referral centres (tertiary centres with a specific gynaecological oncology unit) or other types of centre. Participants Consecutive adult patients presenting with an adnexal mass between January 2012 and March 2015 and managed by surgery or follow-up. Main outcome measures Overall and centre specific discrimination, calibration, and clinical utility of six prediction models for ovarian malignancy (risk of malignancy index (RMI), logistic regression model 2 (LR2), simple rules, simple rules risk model (SRRisk), assessment of different neoplasias in the adnexa (ADNEX) with or without CA125). ADNEX allows the risk of malignancy to be subdivided into risks of a borderline, stage I primary, stage II-IV primary, or secondary metastatic malignancy. The outcome was based on histology if patients underwent surgery, or on results of clinical and ultrasound follow-up at 12 (±2) months. Multiple imputation was used when outcome based on follow-up was uncertain. Results The primary analysis included 17 centres that met strict quality criteria for surgical and follow-up data (5717 of all 8519 patients). 812 patients (14%) had a mass that was already in follow-up at study recruitment, therefore 4905 patients were included in the statistical analysis. The outcome was benign in 3441 (70%) patients and malignant in 978 (20%). Uncertain outcomes (486, 10%) were most often explained by limited follow-up information. The overall area under the receiver operating characteristic curve was highest for ADNEX with CA125 (0.94, 95% confidence interval 0.92 to 0.96), ADNEX without CA125 (0.94, 0.91 to 0.95) and SRRisk (0.94, 0.91 to 0.95), and lowest for RMI (0.89, 0.85 to 0.92). Calibration varied among centres for all models, however the ADNEX models and SRRisk were the best calibrated. Calibration of the estimated risks for the tumour subtypes was good for ADNEX irrespective of whether or not CA125 was included as a predictor. Overall clinical utility (net benefit) was highest for the ADNEX models and SRRisk, and lowest for RMI. For patients who received at least one follow-up scan (n=1958), overall area under the receiver operating characteristic curve ranged from 0.76 (95% confidence interval 0.66 to 0.84) for RMI to 0.89 (0.81 to 0.94) for ADNEX with CA125. Conclusions Our study found the ADNEX models and SRRisk are the best models to distinguish between benign and malignant masses in all patients presenting with an adnexal mass, including those managed conservatively. Trial registration ClinicalTrials.gov NCT01698632.

2 January 2025 in Nature Medicine
Verified
Journal Article
Open access
PDF file
FCFilip Christiansen
Filip Christiansen
EKEmir Konuk
Emir Konuk
AGAdithya Raju Ganeshan
Adithya Raju Ganeshan
RWRobert Welch
Robert Welch
JHJoana Palés Huix
Joana Palés Huix
EEElisabeth Epstein
Elisabeth Epstein
Ovarian lesions are common and often incidentally detected. A critical shortage of expert ultrasound examiners has raised concerns of unnecessary interventions and delayed cancer diagnoses. Deep learning has shown promising results in the detection of ovarian cancer in ultrasound images; however, external validation is lacking. In this international multicenter retrospective study, we developed and validated transformer-based neural network models using a comprehensive dataset of 17,119 ultrasound images from 3,652 patients across 20 centers in eight countries. Using a leave-one-center-out cross-validation scheme, for each center in turn, we trained a model using data from the remaining centers. The models demonstrated robust performance across centers, ultrasound systems, histological diagnoses and patient age groups, significantly outperforming both expert and non-expert examiners on all evaluated metrics, namely F1 score, sensitivity, specificity, accuracy, Cohen’s kappa, Matthew’s correlation coefficient, diagnostic odds ratio and Youden’s J statistic. Furthermore, in a retrospective triage simulation, artificial intelligence (AI)-driven diagnostic support reduced referrals to experts by 63% while significantly surpassing the diagnostic performance of the current practice. These results show that transformer-based models exhibit strong generalization and above human expert-level diagnostic accuracy, with the potential to alleviate the shortage of expert ultrasound examiners and improve patient outcomes.

4 December 2018 in Cancer Imaging
Verified
Journal Article
Open access
PDF file
SRStefania Rizzo
Stefania Rizzo
MFMarco Femia
Marco Femia
VBValentina Buscarino
Valentina Buscarino
DFDorella Franchi
Dorella Franchi
AGAnnalisa Garbi
Annalisa Garbi
MBMassimo Bellomi
Massimo Bellomi
Endometrial cancer is the most common gynaecologic malignancy in developed countries and its incidence is increasing. First-level treatment, if no contraindicated, is based on surgery. Pre-operative imaging is needed for evaluation of local extent and detection of distant metastases in order to guide treatment planning. Radiological evaluation, based on transvaginal ultrasound, MR and CT, can make the difference in disease management, paying special attention to assessment of entity of myometrial invasion, cervical stromal extension, and assessment of lymph nodal involvement and distant metastases.

2 June 2017 in Diagnostics
Verified
Journal Article
Open access
PDF file
WFWouter Froyman
Wouter Froyman
LWLaure Wynants
Laure Wynants
CLChiara Landolfo
Chiara Landolfo
TBTom Bourne
Tom Bourne
LVLil Valentin
Lil Valentin
DTDirk Timmerman
Dirk Timmerman
Background: The aim of this study was to assess and compare the performance of different ultrasound-based International Ovarian Tumor Analysis (IOTA) strategies and subjective assessment for the diagnosis of early stage ovarian malignancy. Methods: This is a secondary analysis of a prospective multicenter cross-sectional diagnostic accuracy study that included 1653 patients recruited at 18 centers from 2009 to 2012. All patients underwent standardized transvaginal ultrasonography by experienced ultrasound investigators. We assessed test performance of the IOTA Simple Rules (SRs), Simple Rules Risk (SRR), the Assessment of Different NEoplasias in the adneXa (ADNEX) model and subjective assessment to discriminate between stage I-II ovarian cancer and benign disease. Reference standard was histology after surgery. Results: 230 (13.9%) patients proved to have stage I–II primary invasive ovarian malignancy, and 1423 (86.1%) had benign disease. Sensitivity and specificity with respect to malignancy (95% confidence intervals) of the original SRs (classifying all inconclusive cases as malignant) were 94.3% (90.6% to 96.7%) and 73.4% (71.0% to 75.6%). Subjective assessment had a sensitivity and specificity of 90.0% (85.4% to 93.2%) and 86.7% (84.9% to 88.4%), respectively. The areas under the receiver operator characteristic curves of SRR and ADNEX were 0.917 (0.902 to 0.933) and 0.905 (0.920 to 0.934), respectively. At a 1% risk cut-off, sensitivity and specificity for SRR were 100% (98.4% to 100%) and 38.0% (35.5% to 40.6%), and for ADNEX were 100% (98.4% to 100%) and 19.4% (17.4% to 21.5%). At a 30% risk cut-off, sensitivity and specificity for SRR were 88.3% (83.5% to 91.8%) and 81.1% (79% to 83%), and for ADNEX were 84.5% (80.5% to 89.6%) and 84.5% (82.6% to 86.3%). Conclusion: This study shows that all three IOTA strategies have good ability to discriminate between stage I-II ovarian malignancy and benign disease.

1 August 2012 in Ultrasound in Obstetrics & Gynecology
Verified
Journal Article
Open access
PDF file
DFD. Franchi
D. Franchi
SBSara Boveri
Sara Boveri
RFRobert Fruscio
Robert Fruscio
DFDaniela Fischerova
Daniela Fischerova
SGStefano Guerriero
Stefano Guerriero
ATA. C. Testa
A. C. Testa
Objectives: To describe the sonographic characteristics of borderline ovarian tumor (BOT) recurrence.Methods: From the databases of five ultrasound centers, we retrospectively identified 68 patients with histological diagnosis of recurrent BOT who had undergone preoperative ultrasound examination. All recurrences were detected during planned follow‐up ultrasound examinations. Recurrent lesions were described using the terms and definitions of the International Ovarian Tumor Analysis (IOTA) group.Results: Sixty‐two patients had a serous BOT recurrence and six a mucinous BOT recurrence. All patients except one were premenopausal, 84% of them being < 40 years old. All but one patient were asymptomatic at diagnosis of the recurrence. Fertility‐sparing surgery of the recurrent tumor was performed in 57/68 (84%) patients. The most frequent ultrasound feature of recurrent serous BOT was a unilocular solid cyst (49/62, 79%) and almost half of the recurrent serous BOTs (29/62, 47%) had multiple papillary projections. In 89% of the recurrent serous BOTs there was at least one papillation with irregular surface and in 73% there was at least one papillation vascularized at color Doppler examination. Recurrent mucinous BOTs appeared mainly as multilocular or multilocular solid cysts (5/6, 83%).Conclusion: Sonographic features of recurrent BOT resemble those described by others for different subtypes of primary BOT. Copyright © 2012 ISUOG. Published by John Wiley & Sons, Ltd.

3 August 2018 in Ultrasound in Obstetrics & Gynecology
Verified
Journal Article
Open access
CLC. Landolfo
C. Landolfo
LVLil Valentin
Lil Valentin
DFD. Franchi
D. Franchi
CHC. Van Holsbeke
C. Van Holsbeke
RFRobert Fruscio
Robert Fruscio
DTD. Timmerman
D. Timmerman
Objectives: To identify ultrasound features of papillations or of the cyst wall that can discriminate between benign and malignant unilocular‐solid cysts with papillations but no other solid components.Methods: From the International Ovarian Tumor Analysis (IOTA) database derived from seven ultrasound centers, we identified patients with an adnexal lesion described at ultrasonography as unilocular‐solid with papillations but no other solid components. All patients had undergone transvaginal ultrasound between 1999 and 2007 or 2009 and 2012, by an experienced examiner following the IOTA research protocol. Information on four ultrasound features of papillations had been collected prospectively. Information on a further seven ultrasound features was collected retrospectively from electronic or paper ultrasound images of good quality. The histological diagnosis of the surgically removed adnexal lesion was considered the gold standard.Results: Of 204 masses included, 131 (64.2%) were benign, 42 (20.6%) were borderline tumors, 30 (14.7%) were primary invasive tumors and one (0.5%) was a metastasis. Multivariate logistic regression analysis showed the following ultrasound features to be associated independently with malignancy: height of the largest papillation, presence of blood flow in papillations, papillation confluence or dissemination, and shadows behind papillations. Shadows decreased the odds of malignancy, while the other features increased them.Conclusion: We have identified ultrasound features that can help to discriminate between benign and malignant unilocular‐solid cysts with papillations but no other solid components. Our results need to be confirmed in prospective studies. Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd.