PHAIR Consortium

The Pancreatobiliary and Hepatic Artificial Intelligence Research (PHAIR) Consortium. Focusing on clinically relevant and applicable AI tools voor HPB disease.

[Giovanni Marchegiani @Gio_Marchegiani]

🎄 Pancreas 2023 Wrapped! 🍾

🔟 October 2023: The use of radiomics 🤖 in pancreas surgery... future or present?!

🧐 Is radiomics already impacting your clinical practice? 🫵

This is the first tool to predict the risk of POPF 🐳

[Giovanni Marchegiani @Gio_Marchegiani]

The first ever Radiomics 👾 preoperative fistula risk score !!!

Development 🇳🇱 and external validation 🇮🇹

🐳 The radiomics-based preop FRS performed similarly to the existing scores ( ROC 0.8)

💪 Great added value considering only CT derived data!

https://t.co/E6Brd9WDhv

[PHAIR Consortium @PHAIRconsortium]

Now in @BjsOpen!

Externally valid #radiomics pre-op fistula risk score (RAD-FRS) for Whipple! 🇮🇹🇳🇱

FIRST externally valid model ✅ Similar AUC (0.81) to (ua)FRS (0.79) in preop setting 📷

👏 @EwoutIngwersen, @JackieBereska and @PHAIRconsortium!

https://t.co/7jBtcfpGLR

[PHAIR Consortium @PHAIRconsortium]

Last week radiogist Håvard Bjørke Jenssen from 🇳🇴@Oslounivsykehus visited the 🇳🇱 @AmsterdamUMC HPB team!

Sharing and aligning HPB CT segmentation ✍️ strategies with Yung Nio, @JackieBereska @i_verpalen and others! 📸

Visit finalized with the annual @AmsterdamUMC HPB BBQ!

[PHAIR Consortium @PHAIRconsortium]

Last week radiogist Håvard Bjørke Jenssen from 🇳🇴@Oslounivsykehus visited the 🇳🇱 @AmsterdamUMC HPB team!

Sharing and aligning HPB CT segmentation ✍️ strategies with Yung Nio, @JackieBereska @i_verpalen and others! 📸

Visit finalized with the annual @AmsterdamUMC HPB BBQ!

[PHAIR Consortium @PHAIRconsortium]

This week 🇳🇱 @i_verpalen @JackieBereska etc. from @amsterdamumc visit 🇳🇴@AAFretland @knutjla in the @Oslounivsykehus!

External validation of a CRLM CT segmentation model! 📸

Discussions on imaging-based response evaluation to neoadjuvant therapy in both liver and pancreas! 💉

[PHAIR Consortium @PHAIRconsortium]

This week 🇳🇱 @i_verpalen @JackieBereska etc. from @amsterdamumc visit 🇳🇴@AAFretland @knutjla in the @Oslounivsykehus!

External validation of a CRLM CT segmentation model! 📸

Discussions on imaging-based response evaluation to neoadjuvant therapy in both liver and pancreas! 💉

[PHAIR Consortium @PHAIRconsortium]

📢: @albe_balduzzi @BorisJanssen and @PHAIRconsortium in @BJSurgery

📈 12 imaging-based AI models to predict malignancy of IPMN

📊 Methodologically suboptimal

🔎 Methodologically robust prospective studies incl. non-surgical patients required!

Read: https://t.co/LPMOGfUnD8

[PHAIR Consortium @PHAIRconsortium]

📢: @albe_balduzzi @BorisJanssen and @PHAIRconsortium in @BJSurgery

📈 12 imaging-based AI models to predict malignancy of IPMN

📊 Methodologically suboptimal

🔎 Methodologically robust prospective studies incl. non-surgical patients required!

Read: https://t.co/LPMOGfUnD8

[Giovanni Marchegiani @Gio_Marchegiani]

The future of IPMN management 🍇 will be based on two pillars

🏛️ Biomarkers and AI

🤖 Twelve AI based models to assess the risk of malignancy already exist!

😢 Methodological quality and reliability still to be improved… let’s work on it!!

🆓 read 👉https://t.co/U84qBoSA7R

[PHAIR Consortium @PHAIRconsortium]

RT @MarcBesselink: Can #AI based models predict the risk of #PancreaticCancer in #IPMN?
⬇️
Scoping review @BJSurgery @PHAIRconsortium op…

[PHAIR Consortium @PHAIRconsortium]

@EurPancClub #EPC2023

@BorisJanssen on behalf of @PHAIRconsortium!

AI residual cancer burden (RCB) to evaluate impact of NAT in resected pancreatic cancer!

RCB 🔬
NAT vs. upfront: 2.2 vs. 10.3 mm2
p<0.0001

Survival 📈
Above vs. below median RCB: 24 vs. 30 months.
p=0.018

[PHAIR Consortium @PHAIRconsortium]

@EurPancClub #EPC2023

@BorisJanssen on behalf of @PHAIRconsortium!

AI residual cancer burden (RCB) to evaluate impact of NAT in resected pancreatic cancer!

RCB 🔬
NAT vs. upfront: 2.2 vs. 10.3 mm2
p<0.0001

Survival 📈
Above vs. below median RCB: 24 vs. 30 months.
p=0.018

[PHAIR Consortium @PHAIRconsortium]

@EsgarSociety #esgar2023

@PHAIRconsortium lead @i_verpalen presents:

🎙️ #ArtificialIntelligence based Assessment of Vascular Involvement and Resectability 🔪on CT in patients with PDAC

Segmentation 📸 below!

Great work @JackieBereska @BorisJanssen and @PHAIRconsortium ❗️

[PHAIR Consortium @PHAIRconsortium]

@EsgarSociety #esgar2023

@PHAIRconsortium lead @i_verpalen presents:

🎙️ #ArtificialIntelligence based Assessment of Vascular Involvement and Resectability 🔪on CT in patients with PDAC

Segmentation 📸 below!

Great work @JackieBereska @BorisJanssen and @PHAIRconsortium ❗️

[PHAIR Consortium @PHAIRconsortium]

Now @EsgarSociety #ESGAR2023

Externally valid #radiomics preoperative fistula risk score (RAD-FRS) for Whipple!

FIRST externally valid model of its kind ✅
Similar AUC (0.81) to (ua)FRS (0.79) in preop setting ✅

Great work @EwoutIngwersen,
@JackieBereska and @PHAIRconsortium!

[PHAIR Consortium @PHAIRconsortium]

Now @EsgarSociety #ESGAR2023

Externally valid #radiomics preoperative fistula risk score (RAD-FRS) for Whipple!

FIRST externally valid model of its kind ✅
Similar AUC (0.81) to (ua)FRS (0.79) in preop setting ✅

Great work @EwoutIngwersen,
@JackieBereska and @PHAIRconsortium!

[PHAIR Consortium @PHAIRconsortium]

#EAHPBA2023 @EAHPBA

Segmentation model for 🔬 residual pancreatic cancer following neoadjuvant therapy (ISGPP-2)!

Model based on 528 pt, 14 centers, 7 countries🇳🇱🇺🇸🇩🇪🇳🇴🇹🇷🇰🇷🇦🇺, 4 scanner types!

Validated with by-scanner internal-external cross-validation 📊! F1 score of 0.78!

[PHAIR Consortium @PHAIRconsortium]

#EAHPBA2023 @EAHPBA

External validation of #radiomics preoperative fistula💧 risk score (RAD-FRS) for Whipple 🔪 !

FIRST externally validated model of its kind ✅
Similar AUC to (ua)FRS in pre-op setting ✅

Great work by @EwoutIngwersen, @JackieBereska and @PHAIRconsortium!

[PHAIR Consortium @PHAIRconsortium]

#EAHPBA2023 @EAHPBA

Segmentation model for 🔬 residual pancreatic cancer following neoadjuvant therapy (ISGPP-2)!

Model based on 528 pt, 14 centers, 7 countries🇳🇱🇺🇸🇩🇪🇳🇴🇹🇷🇰🇷🇦🇺, 4 scanner types!

Validated with by-scanner internal-external cross-validation 📊! F1 score of 0.78!